Complete pipeline

README.md

@@ -1,3 +1,39 @@
 # AutoKG
 
-Automatic knowledge graph construction using ChatGPT.
+Using large language models (ChatGPT) to automatically construct a knowledge graph from unstructured plain text.
+
+## Usage
+
+```
+python text2kg.py <infile> [-h] [--output OUTPUT] [--cookbook COOKBOOK] [--recipe RECIPE] [--thoughts]
+```
+
+**Parameters**
+
+```
+infile
+    path to input text file (assumed to be in plain text form)
+--output
+    directory to save results to (default: ./output)
+--cookbook
+    path to JSON file containing GPT prompts (default: ./recipes.json)
+--recipe
+    recipe to execute, must be a valid name from cookbook (see above)
+--thoughts
+    optional flag to save intermediary GPT prompts/replies
+```
+
+## File structure
+
+### [`data`](./data/)
+
+Name | Description | Source
+--- | --- | ---
+`data/openstax/OS_bio_4_2.txt` | Chapter 4.2 of the OpenStax Biology 2nd edition textbook. | 
+`data/Seq2KG/...` | Processed test datasets from Seq2KG paper. | [GitHub](https://github.com/Michael-Stewart-Webdev/Seq2KG/tree/master)
+
+## References
+
+1. A case study in bootstrapping ontology graphs from textbooks (V. K. Chaudhri et al., 2021)
+2. Seq2KG: an end-to-end neural model for domain agnostic knowledge graph (not text graph) construction from text (M. Stewart & W. Liu, 2020)
+3. Language models are open knowledge graphs (C. Wang et al., 2020)

data/openstax/bio2e/OS_bio_4_2.txt

@@ -0,0 +1,66 @@
+A cell is the smallest unit of a living thing.
+Whether comprised of one cell (like bacteria) or many cells (like a human), we call it an organism.
+Thus, cells are the basic building blocks of all organisms.
+Several cells of one kind that interconnect with each other and perform a shared function form tissues.
+These tissues combine to form an organ (your stomach, heart, or brain), and several organs comprise an organ system (such as the digestive system, circulatory system, or nervous system).
+Several systems that function together form an organism (like a human being).
+Here, we will examine the structure and function of cells.
+There are many types of cells, which scientists group into one of two broad categories: prokaryotic and eukaryotic.
+For example, we classify both animal and plant cells as eukaryotic cells; whereas, we classify bacterial cells as prokaryotic.
+Before discussing the criteria for determining whether a cell is prokaryotic or eukaryotic, we will first examine how biologists study cells.
+Cells vary in size.
+With few exceptions, we cannot see individual cells with the naked eye, so scientists use microscopes (micro- = "small"; -scope = "to look at") to study them.
+A microscope is an instrument that magnifies an object.
+We photograph most cells with a microscope, so we can call these images micrographs.
+The optics of a microscope's lenses change the image orientation that the user sees.
+A specimen that is right-side up and facing right on the microscope slide will appear upside-down and facing left when one views through a microscope, and vice versa.
+Similarly, if one moves the slide left while looking through the microscope, it will appear to move right, and if one moves it down, it will seem to move up.
+This occurs because microscopes use two sets of lenses to magnify the image.
+Because of the manner by which light travels through the lenses, this two lens system produces an inverted image (binocular, or dissecting microscopes, work in a similar manner, but include an additional magnification system that makes the final image appear to be upright).
+To give you a sense of cell size, a typical human red blood cell is about eight millionths of a meter or eight micrometers (abbreviated as eight μm) in diameter.
+A pin head is about two thousandths of a meter (two mm) in diameter.
+That means about 250 red blood cells could fit on a pinhead.
+Most student microscopes are light microscopes (Figure 4.2a).
+Visible light passes and bends through the lens system to enable the user to see the specimen.
+Light microscopes are advantageous for viewing living organisms, but since individual cells are generally transparent, their components are not distinguishable unless they are colored with special stains.
+Staining, however, usually kills the cells.
+Light microscopes that undergraduates commonly use in the laboratory magnify up to approximately 400 times.
+Two parameters that are important in microscopy are magnification and resolving power.
+Magnification is the process of enlarging an object in appearance.
+Resolving power is the microscope's ability to distinguish two adjacent structures as separate: the higher the resolution, the better the image's clarity and detail.
+When one uses oil immersion lenses to study small objects, magnification usually increases to 1,000 times.
+In order to gain a better understanding of cellular structure and function, scientists typically use electron microscopes.
+a) Most light microscopes in a college biology lab can magnify cells up to approximately 400 times and have a resolution of about 200 nanometers.
+b) Electron microscopes provide a much higher magnification, 100,000x, and a have a resolution of 50 picometers.
+use a beam of electrons instead of a beam of light.
+Not only does this allow for higher magnification and, thus, more detail (Figure 4.3), it also provides higher resolving power.
+The method to prepare the specimen for viewing with an electron microscope kills the specimen.
+Electrons have short wavelengths (shorter than photons) that move best in a vacuum, so we cannot view living cells with an electron microscope.
+In a scanning electron microscope, a beam of electrons moves back and forth across a cell's surface, creating details of cell surface characteristics.
+In a transmission electron microscope, the electron beam penetrates the cell and provides details of a cell's internal structures.
+As you might imagine, electron microscopes are significantly more bulky and expensive than light microscopes.
+a) These Salmonella bacteria appear as tiny purple dots when viewed with a light microscope.
+b) This scanning electron microscope micrograph shows Salmonella bacteria (in red) invading human cells (yellow).
+Even though subfigure (b) shows a different Salmonella specimen than subfigure (a), you can still observe the comparative increase in magnification and detail.
+For another perspective on cell size, try the HowBig interactive at this site.
+The microscopes we use today are far more complex than those that Dutch shopkeeper Antony van Leeuwenhoek, used in the 1600s.
+Skilled in crafting lenses, van Leeuwenhoek observed the movements of single-celled organisms, which he collectively termed "animalcules."
+In the 1665 publication Micrographia, experimental scientist Robert Hooke coined the term "cell" for the box-like structures he observed when viewing cork tissue through a lens.
+In the 1670s, van Leeuwenhoek discovered bacteria and protozoa.
+Later advances in lenses, microscope construction, and staining techniques enabled other scientists to see some components inside cells.
+By the late 1830s, botanist Matthias Schleiden and zoologist Theodor Schwann were studying tissues and proposed the unified cell theory, which states that one or more cells comprise all living things, the cell is the basic unit of life, and new cells arise from existing cells.
+Rudolf Virchow later made important contributions to this theory.
+Have you ever heard of a medical test called a Pap smear (Figure 4.4)?
+In this test, a doctor takes a small sample of cells from the patient's uterine cervix and sends it to a medical lab where a cytotechnologist stains the cells and examines them for any changes that could indicate cervical cancer or a microbial infection.
+Cytotechnologists (cyto- = "cell") are professionals who study cells via microscopic examinations and other laboratory tests.
+They are trained to determine which cellular changes are within normal limits and which are abnormal.
+Their focus is not limited to cervical cells.
+They study cellular specimens that come from all organs.
+When they notice abnormalities, they consult a pathologist, a medical doctor who interprets and diagnoses changes that disease in body tissue and fluids cause.
+Cytotechnologists play a vital role in saving people's lives.
+When doctors discover abnormalities early, a patient's treatment can begin sooner, which usually increases the chances of a successful outcome.
+These uterine cervix cells, viewed through a light microscope, are from a Pap smear.
+Normal cells are on the left.
+The cells on the right are infected with human papillomavirus (HPV).
+Notice that the infected cells are larger.
+Also, two of these cells each have two nuclei instead of one, the normal number.

output/2023-08-28_triplets_Direct.json

@@ -0,0 +1,526 @@
+[
+    {
+        "sentence": "A cell is the smallest unit of a living thing.",
+        "triplets": [
+            "cell::is the smallest unit of::a living thing"
+        ]
+    },
+    {
+        "sentence": "Whether comprised of one cell (like bacteria) or many cells (like a human), we call it an organism.",
+        "triplets": [
+            "organism::comprised of::one cell",
+            "organism::comprised of::many cells",
+            "bacteria::comprised of::one cell",
+            "human::comprised of::many cells"
+        ]
+    },
+    {
+        "sentence": "Thus, cells are the basic building blocks of all organisms.",
+        "triplets": [
+            "cells::are the basic building blocks of::all organisms"
+        ]
+    },
+    {
+        "sentence": "Several cells of one kind that interconnect with each other and perform a shared function form tissues.",
+        "triplets": [
+            "cells::interconnect with::each other",
+            "cells::perform::shared function",
+            "cells::form::tissues"
+        ]
+    },
+    {
+        "sentence": "These tissues combine to form an organ (your stomach, heart, or brain), and several organs comprise an organ system (such as the digestive system, circulatory system, or nervous system).",
+        "triplets": [
+            "tissues::combine to form::organ",
+            "organ::comprise::organ system",
+            "organ system::such as::digestive system",
+            "organ system::such as::circulatory system",
+            "organ system::such as::nervous system"
+        ]
+    },
+    {
+        "sentence": "Several systems that function together form an organism (like a human being).",
+        "triplets": [
+            "systems::form::organism",
+            "systems::function together::organism",
+            "systems::form::human being"
+        ]
+    },
+    {
+        "sentence": "Here, we will examine the structure and function of cells.",
+        "triplets": [
+            "Here::examine::structure and function of cells"
+        ]
+    },
+    {
+        "sentence": "There are many types of cells, which scientists group into one of two broad categories: prokaryotic and eukaryotic.",
+        "triplets": [
+            "cells::have types::many",
+            "scientists::group into::categories",
+            "types::group into::categories",
+            "prokaryotic cells::belong to::categories",
+            "eukaryotic cells::belong to::categories"
+        ]
+    },
+    {
+        "sentence": "For example, we classify both animal and plant cells as eukaryotic cells; whereas, we classify bacterial cells as prokaryotic.",
+        "triplets": [
+            "animal cells::classified as::eukaryotic cells",
+            "plant cells::classified as::eukaryotic cells",
+            "bacterial cells::classified as::prokaryotic"
+        ]
+    },
+    {
+        "sentence": "Before discussing the criteria for determining whether a cell is prokaryotic or eukaryotic, we will first examine how biologists study cells.",
+        "triplets": [
+            "biologists::study::cells",
+            "cell::is::prokaryotic",
+            "cell::is::eukaryotic",
+            "criteria::determining::cell",
+            "cell::is::prokaryotic",
+            "cell::is::eukaryotic"
+        ]
+    },
+    {
+        "sentence": "Cells vary in size.",
+        "triplets": [
+            "Cells::vary in::size"
+        ]
+    },
+    {
+        "sentence": "With few exceptions, we cannot see individual cells with the naked eye, so scientists use microscopes (micro- = \"small\"; -scope = \"to look at\") to study them.",
+        "triplets": [
+            "scientists::use::microscopes",
+            "microscopes::study::cells",
+            "cells::cannot see with::naked eye",
+            "scientists::use::microscopes",
+            "microscopes::study::them"
+        ]
+    },
+    {
+        "sentence": "A microscope is an instrument that magnifies an object.",
+        "triplets": []
+    },
+    {
+        "sentence": "We photograph most cells with a microscope, so we can call these images micrographs.",
+        "triplets": [
+            "We::photograph::most cells",
+            "We::call::these images"
+        ]
+    },
+    {
+        "sentence": "The optics of a microscope's lenses change the image orientation that the user sees.",
+        "triplets": [
+            "microscope's lenses::change::image orientation",
+            "optics::change::image orientation",
+            "user::sees::image orientation"
+        ]
+    },
+    {
+        "sentence": "A specimen that is right-side up and facing right on the microscope slide will appear upside-down and facing left when one views through a microscope, and vice versa.",
+        "triplets": [
+            "specimen::is right-side up and facing right on::microscope slide",
+            "specimen::will appear upside-down and facing left when one views through::microscope",
+            "microscope slide::is viewed through::microscope",
+            "microscope::views through::microscope slide"
+        ]
+    },
+    {
+        "sentence": "Similarly, if one moves the slide left while looking through the microscope, it will appear to move right, and if one moves it down, it will seem to move up.",
+        "triplets": [
+            "one::moves::slide",
+            "slide::left::microscope",
+            "one::moves::slide",
+            "slide::right::appear",
+            "one::moves::slide",
+            "slide::down::microscope",
+            "one::moves::slide",
+            "slide::up::appear"
+        ]
+    },
+    {
+        "sentence": "This occurs because microscopes use two sets of lenses to magnify the image.",
+        "triplets": [
+            "This occurs::because::microscopes",
+            "microscopes::use::two sets of lenses",
+            "two sets of lenses::magnify::the image"
+        ]
+    },
+    {
+        "sentence": "Because of the manner by which light travels through the lenses, this two lens system produces an inverted image (binocular, or dissecting microscopes, work in a similar manner, but include an additional magnification system that makes the final image appear to be upright).",
+        "triplets": [
+            "lenses::travels through::light",
+            "two lens system::produces::inverted image",
+            "dissecting microscopes::work in::similar manner",
+            "dissecting microscopes::include::additional magnification system",
+            "additional magnification system::makes appear::final image",
+            "final image::appear to be::upright"
+        ]
+    },
+    {
+        "sentence": "To give you a sense of cell size, a typical human red blood cell is about eight millionths of a meter or eight micrometers (abbreviated as eight \u03bcm) in diameter.",
+        "triplets": [
+            "human red blood cell::has size::eight millionths of a meter",
+            "human red blood cell::has size::eight micrometers",
+            "eight micrometers::abbreviated as::eight \u03bcm",
+            "eight \u03bcm::is a unit of::diameter"
+        ]
+    },
+    {
+        "sentence": "A pin head is about two thousandths of a meter (two mm) in diameter.",
+        "triplets": []
+    },
+    {
+        "sentence": "That means about 250 red blood cells could fit on a pinhead.",
+        "triplets": [
+            "red blood cells::fit on::pinhead",
+            "250::fit on::pinhead",
+            "red blood cells::could::fit on",
+            "250::could::fit on",
+            "That::means::about"
+        ]
+    },
+    {
+        "sentence": "Most student microscopes are light microscopes (Figure 4.2a).",
+        "triplets": [
+            "student microscopes::are::light microscopes",
+            "Figure 4.2a::is::light microscopes"
+        ]
+    },
+    {
+        "sentence": "Visible light passes and bends through the lens system to enable the user to see the specimen.",
+        "triplets": [
+            "Visible light::passes through::lens system",
+            "Visible light::bends through::lens system",
+            "lens system::enables::user",
+            "user::sees::specimen"
+        ]
+    },
+    {
+        "sentence": "Light microscopes are advantageous for viewing living organisms, but since individual cells are generally transparent, their components are not distinguishable unless they are colored with special stains.",
+        "triplets": [
+            "Light microscopes::are advantageous for::viewing living organisms",
+            "individual cells::are generally transparent::not distinguishable",
+            "individual cells::are generally transparent::colored with special stains"
+        ]
+    },
+    {
+        "sentence": "Staining, however, usually kills the cells.",
+        "triplets": [
+            "Staining::kills::cells"
+        ]
+    },
+    {
+        "sentence": "Light microscopes that undergraduates commonly use in the laboratory magnify up to approximately 400 times.",
+        "triplets": [
+            "Light microscopes::use::undergraduates",
+            "undergraduates::use::laboratory",
+            "Light microscopes::magnify::up to approximately 400 times"
+        ]
+    },
+    {
+        "sentence": "Two parameters that are important in microscopy are magnification and resolving power.",
+        "triplets": [
+            "microscopy::has parameter::magnification",
+            "microscopy::has parameter::resolving power"
+        ]
+    },
+    {
+        "sentence": "Magnification is the process of enlarging an object in appearance.",
+        "triplets": [
+            "Magnification::is the process of::enlarging an object in appearance"
+        ]
+    },
+    {
+        "sentence": "Resolving power is the microscope's ability to distinguish two adjacent structures as separate: the higher the resolution, the better the image's clarity and detail.",
+        "triplets": [
+            "Resolving power::is::microscope's ability",
+            "microscope's ability::to distinguish::two adjacent structures",
+            "two adjacent structures::as separate::the higher the resolution",
+            "the higher the resolution::the better::image's clarity and detail"
+        ]
+    },
+    {
+        "sentence": "When one uses oil immersion lenses to study small objects, magnification usually increases to 1,000 times.",
+        "triplets": [
+            "oil immersion lenses::used to study::small objects",
+            "magnification::increases to::1"
+        ]
+    },
+    {
+        "sentence": "In order to gain a better understanding of cellular structure and function, scientists typically use electron microscopes.",
+        "triplets": [
+            "scientists::use::electron microscopes",
+            "scientists::gain::better understanding",
+            "cellular structure and function::understanding of::scientists",
+            "electron microscopes::used by::scientists"
+        ]
+    },
+    {
+        "sentence": "a) Most light microscopes in a college biology lab can magnify cells up to approximately 400 times and have a resolution of about 200 nanometers.",
+        "triplets": [
+            "light microscopes::in::college biology lab",
+            "light microscopes::magnify::cells",
+            "light microscopes::up to::approximately 400 times",
+            "light microscopes::have::resolution",
+            "light microscopes::of::about 200 nanometers"
+        ]
+    },
+    {
+        "sentence": "b) Electron microscopes provide a much higher magnification, 100,000x, and a have a resolution of 50 picometers.",
+        "triplets": [
+            "Electron microscopes::provide::higher magnification",
+            "Electron microscopes::have::resolution",
+            "Electron microscopes::have::50 picometers"
+        ]
+    },
+    {
+        "sentence": "use a beam of electrons instead of a beam of light.",
+        "triplets": [
+            "beam::use::electrons",
+            "beam::instead of::light"
+        ]
+    },
+    {
+        "sentence": "Not only does this allow for higher magnification and, thus, more detail (Figure 4.3), it also provides higher resolving power.",
+        "triplets": [
+            "higher magnification::allow for::more detail",
+            "Figure 4.3::provides::higher resolving power"
+        ]
+    },
+    {
+        "sentence": "The method to prepare the specimen for viewing with an electron microscope kills the specimen.",
+        "triplets": [
+            "The method::to prepare::the specimen",
+            "The specimen::for viewing with::an electron microscope",
+            "The method::kills::the specimen"
+        ]
+    },
+    {
+        "sentence": "Electrons have short wavelengths (shorter than photons) that move best in a vacuum, so we cannot view living cells with an electron microscope.",
+        "triplets": [
+            "Electrons::have::short wavelengths",
+            "Electrons::move best in::a vacuum",
+            "we::cannot view::living cells",
+            "an electron microscope::view::living cells"
+        ]
+    },
+    {
+        "sentence": "In a scanning electron microscope, a beam of electrons moves back and forth across a cell's surface, creating details of cell surface characteristics.",
+        "triplets": [
+            "scanning electron microscope::has::beam of electrons",
+            "beam of electrons::moves back and forth across::cell's surface",
+            "beam of electrons::creates::details of cell surface characteristics"
+        ]
+    },
+    {
+        "sentence": "In a transmission electron microscope, the electron beam penetrates the cell and provides details of a cell's internal structures.",
+        "triplets": [
+            "transmission electron microscope::has::electron beam",
+            "electron beam::penetrates::cell",
+            "electron beam::provides details of::cell's internal structures"
+        ]
+    },
+    {
+        "sentence": "As you might imagine, electron microscopes are significantly more bulky and expensive than light microscopes.",
+        "triplets": [
+            "electron microscopes::are significantly more bulky and expensive than::light microscopes"
+        ]
+    },
+    {
+        "sentence": "a) These Salmonella bacteria appear as tiny purple dots when viewed with a light microscope.",
+        "triplets": [
+            "Salmonella bacteria::appear as::tiny purple dots",
+            "Salmonella bacteria::viewed with::light microscope"
+        ]
+    },
+    {
+        "sentence": "b) This scanning electron microscope micrograph shows Salmonella bacteria (in red) invading human cells (yellow).",
+        "triplets": [
+            "Salmonella bacteria::invading::human cells"
+        ]
+    },
+    {
+        "sentence": "Even though subfigure (b) shows a different Salmonella specimen than subfigure (a), you can still observe the comparative increase in magnification and detail.",
+        "triplets": [
+            "you::observe::the comparative increase in magnification and detail"
+        ]
+    },
+    {
+        "sentence": "For another perspective on cell size, try the HowBig interactive at this site.",
+        "triplets": [
+            "HowBig interactive::is related to::cell size"
+        ]
+    },
+    {
+        "sentence": "The microscopes we use today are far more complex than those that Dutch shopkeeper Antony van Leeuwenhoek, used in the 1600s.",
+        "triplets": [
+            "Antony van Leeuwenhoek::used::microscopes",
+            "microscopes::are::complex",
+            "microscopes::use::today",
+            "Antony van Leeuwenhoek::used::those",
+            "Antony van Leeuwenhoek::used::microscopes",
+            "microscopes::are::complex",
+            "microscopes::use::today"
+        ]
+    },
+    {
+        "sentence": "Skilled in crafting lenses, van Leeuwenhoek observed the movements of single-celled organisms, which he collectively termed \"animalcules.\"",
+        "triplets": [
+            "van Leeuwenhoek::observed::movements of single-celled organisms",
+            "van Leeuwenhoek::crafting::lenses"
+        ]
+    },
+    {
+        "sentence": "In the 1665 publication Micrographia, experimental scientist Robert Hooke coined the term \"cell\" for the box-like structures he observed when viewing cork tissue through a lens.",
+        "triplets": [
+            "Robert Hooke::coined::term",
+            "Robert Hooke::observed::box-like structures",
+            "Robert Hooke::viewing::cork tissue",
+            "Micrographia::publication::1665",
+            "Robert Hooke::viewing::cork tissue through a lens"
+        ]
+    },
+    {
+        "sentence": "In the 1670s, van Leeuwenhoek discovered bacteria and protozoa.",
+        "triplets": [
+            "van Leeuwenhoek::discovered::bacteria",
+            "van Leeuwenhoek::discovered::protozoa"
+        ]
+    },
+    {
+        "sentence": "Later advances in lenses, microscope construction, and staining techniques enabled other scientists to see some components inside cells.",
+        "triplets": [
+            "Later advances::in::lenses",
+            "Later advances::in::microscope construction",
+            "Later advances::in::staining techniques",
+            "scientists::enabled::other scientists",
+            "scientists::to see::components inside cells"
+        ]
+    },
+    {
+        "sentence": "By the late 1830s, botanist Matthias Schleiden and zoologist Theodor Schwann were studying tissues and proposed the unified cell theory, which states that one or more cells comprise all living things, the cell is the basic unit of life, and new cells arise from existing cells.",
+        "triplets": [
+            "Matthias Schleiden::studying::tissues",
+            "Matthias Schleiden::proposed::unified cell theory",
+            "unified cell theory::states::one or more cells comprise all living things",
+            "unified cell theory::states::the cell is the basic unit of life",
+            "unified cell theory::states::new cells arise from existing cells",
+            "Theodor Schwann::studying::tissues",
+            "Theodor Schwann::proposed::unified cell theory",
+            "unified cell theory::states::one or more cells comprise all living things",
+            "unified cell theory::states::the cell is the basic unit of life",
+            "unified cell theory::states::new cells arise from existing cells"
+        ]
+    },
+    {
+        "sentence": "Rudolf Virchow later made important contributions to this theory.",
+        "triplets": [
+            "Rudolf Virchow::made contributions to::theory"
+        ]
+    },
+    {
+        "sentence": "Have you ever heard of a medical test called a Pap smear (Figure 4.4)?",
+        "triplets": [
+            "you::heard of::medical test",
+            "Pap smear::called::a medical test"
+        ]
+    },
+    {
+        "sentence": "In this test, a doctor takes a small sample of cells from the patient's uterine cervix and sends it to a medical lab where a cytotechnologist stains the cells and examines them for any changes that could indicate cervical cancer or a microbial infection.",
+        "triplets": [
+            "doctor::takes::small sample of cells",
+            "doctor::sends::sample of cells to medical lab",
+            "cytotechnologist::stains::cells",
+            "cytotechnologist::examines::cells",
+            "cells::indicate::cervical cancer",
+            "cells::indicate::microbial infection"
+        ]
+    },
+    {
+        "sentence": "Cytotechnologists (cyto- = \"cell\") are professionals who study cells via microscopic examinations and other laboratory tests.",
+        "triplets": [
+            "Cytotechnologists::study::cells",
+            "Cytotechnologists::use::microscopic examinations",
+            "Cytotechnologists::use::other laboratory tests"
+        ]
+    },
+    {
+        "sentence": "They are trained to determine which cellular changes are within normal limits and which are abnormal.",
+        "triplets": [
+            "They::are trained to determine::which cellular changes are within normal limits",
+            "They::are trained to determine::which cellular changes are abnormal"
+        ]
+    },
+    {
+        "sentence": "Their focus is not limited to cervical cells.",
+        "triplets": [
+            "Their::focus::not limited to cervical cells"
+        ]
+    },
+    {
+        "sentence": "They study cellular specimens that come from all organs.",
+        "triplets": [
+            "cellular specimens::come from::all organs"
+        ]
+    },
+    {
+        "sentence": "When they notice abnormalities, they consult a pathologist, a medical doctor who interprets and diagnoses changes that disease in body tissue and fluids cause.",
+        "triplets": [
+            "they::notice::abnormalities",
+            "they::consult::pathologist",
+            "pathologist::interpret::changes",
+            "pathologist::diagnose::disease",
+            "disease::cause::changes",
+            "disease::in::body tissue",
+            "disease::in::fluids"
+        ]
+    },
+    {
+        "sentence": "Cytotechnologists play a vital role in saving people's lives.",
+        "triplets": [
+            "Cytotechnologists::play a role in::saving people's lives"
+        ]
+    },
+    {
+        "sentence": "When doctors discover abnormalities early, a patient's treatment can begin sooner, which usually increases the chances of a successful outcome.",
+        "triplets": [
+            "doctors::discover::abnormalities",
+            "abnormalities::early::treatment",
+            "patient's treatment::begin::sooner",
+            "treatment::increase::chances"
+        ]
+    },
+    {
+        "sentence": "These uterine cervix cells, viewed through a light microscope, are from a Pap smear.",
+        "triplets": [
+            "uterine cervix cells::viewed through::light microscope",
+            "uterine cervix cells::from::Pap smear"
+        ]
+    },
+    {
+        "sentence": "Normal cells are on the left.",
+        "triplets": [
+            "Normal cells::are on::the left"
+        ]
+    },
+    {
+        "sentence": "The cells on the right are infected with human papillomavirus (HPV).",
+        "triplets": [
+            "cells::infected with::human papillomavirus"
+        ]
+    },
+    {
+        "sentence": "Notice that the infected cells are larger.",
+        "triplets": []
+    },
+    {
+        "sentence": "Also, two of these cells each have two nuclei instead of one, the normal number.",
+        "triplets": [
+            "cells::have::two nuclei",
+            "two of these cells::have::two nuclei",
+            "two of these cells::each have::two nuclei",
+            "two of these cells::have::the normal number"
+        ]
+    }
+]

output/2023-08-28_triplets_LogicBased.json

@@ -0,0 +1,473 @@
+[
+    {
+        "sentence": "A cell is the smallest unit of a living thing.",
+        "triplets": [
+            "cell::is the smallest unit of::living thing"
+        ]
+    },
+    {
+        "sentence": "Whether comprised of one cell (like bacteria) or many cells (like a human), we call it an organism.",
+        "triplets": [
+            "organism::comprised of::one cell",
+            "organism::comprised of::many cells"
+        ]
+    },
+    {
+        "sentence": "Thus, cells are the basic building blocks of all organisms.",
+        "triplets": [
+            "Cells::are the basic building blocks of::all organisms"
+        ]
+    },
+    {
+        "sentence": "Several cells of one kind that interconnect with each other and perform a shared function form tissues.",
+        "triplets": [
+            "cells::interconnect with::each other",
+            "interconnected cells::perform::shared function",
+            "interconnected cells::form::tissues"
+        ]
+    },
+    {
+        "sentence": "These tissues combine to form an organ (your stomach, heart, or brain), and several organs comprise an organ system (such as the digestive system, circulatory system, or nervous system).",
+        "triplets": [
+            "tissues::combine to form::organ",
+            "organ::comprise::organ system",
+            "stomach::is an example of::organ",
+            "heart::is an example of::organ",
+            "brain::is an example of::organ",
+            "digestive system::is an example of::organ system",
+            "circulatory system::is an example of::organ system",
+            "nervous system::is an example of::organ system"
+        ]
+    },
+    {
+        "sentence": "Several systems that function together form an organism (like a human being).",
+        "triplets": [
+            "systems::form::organism",
+            "systems::function together::NULL",
+            "organism::like::human being"
+        ]
+    },
+    {
+        "sentence": "Here, we will examine the structure and function of cells.",
+        "triplets": [
+            "text::examines::structure and function of cells"
+        ]
+    },
+    {
+        "sentence": "There are many types of cells, which scientists group into one of two broad categories: prokaryotic and eukaryotic.",
+        "triplets": [
+            "scientists::group::cells",
+            "cells::group into::prokaryotic and eukaryotic"
+        ]
+    },
+    {
+        "sentence": "For example, we classify both animal and plant cells as eukaryotic cells; whereas, we classify bacterial cells as prokaryotic.",
+        "triplets": [
+            "Animal cells::are classified as::eukaryotic cells",
+            "Plant cells::are classified as::eukaryotic cells",
+            "Bacterial cells::are classified as::prokaryotic"
+        ]
+    },
+    {
+        "sentence": "Before discussing the criteria for determining whether a cell is prokaryotic or eukaryotic, we will first examine how biologists study cells.",
+        "triplets": [
+            "Biologists::study::cells",
+            "criteria::determine::cell is prokaryotic or eukaryotic"
+        ]
+    },
+    {
+        "sentence": "Cells vary in size.",
+        "triplets": [
+            "Cells::vary in size::size"
+        ]
+    },
+    {
+        "sentence": "With few exceptions, we cannot see individual cells with the naked eye, so scientists use microscopes (micro- = \"small\"; -scope = \"to look at\") to study them.",
+        "triplets": [
+            "individual cells::cannot see with::naked eye",
+            "scientists::use::microscopes",
+            "microscopes::study::cells"
+        ]
+    },
+    {
+        "sentence": "A microscope is an instrument that magnifies an object.",
+        "triplets": [
+            "microscope::magnifies::object"
+        ]
+    },
+    {
+        "sentence": "We photograph most cells with a microscope, so we can call these images micrographs.",
+        "triplets": [
+            "We::photograph::most cells with a microscope",
+            "We::call::these images"
+        ]
+    },
+    {
+        "sentence": "The optics of a microscope's lenses change the image orientation that the user sees.",
+        "triplets": [
+            "optics::change::image orientation",
+            "microscope's lenses::change::image orientation",
+            "user::see::image"
+        ]
+    },
+    {
+        "sentence": "A specimen that is right-side up and facing right on the microscope slide will appear upside-down and facing left when one views through a microscope, and vice versa.",
+        "triplets": [
+            "specimen::is positioned::right-side up and facing right on the microscope slide",
+            "specimen::appears::upside-down and facing left when viewed through a microscope",
+            "one::views through::a microscope",
+            "specimen::appears::upside-down and facing left",
+            "one::views through::a microscope"
+        ]
+    },
+    {
+        "sentence": "Similarly, if one moves the slide left while looking through the microscope, it will appear to move right, and if one moves it down, it will seem to move up.",
+        "triplets": [
+            "slide::move::left",
+            "slide::appear to move::right",
+            "slide::move::down",
+            "slide::seem to move::up"
+        ]
+    },
+    {
+        "sentence": "This occurs because microscopes use two sets of lenses to magnify the image.",
+        "triplets": [
+            "Microscopes::use::two sets of lenses",
+            "two sets of lenses::magnify::the image"
+        ]
+    },
+    {
+        "sentence": "Because of the manner by which light travels through the lenses, this two lens system produces an inverted image (binocular, or dissecting microscopes, work in a similar manner, but include an additional magnification system that makes the final image appear to be upright).",
+        "triplets": [
+            "two lens system::produces::inverted image",
+            "binocular or dissecting microscopes::work in::similar manner",
+            "binocular or dissecting microscopes::include::additional magnification system",
+            "additional magnification system::makes::final image appear to be upright"
+        ]
+    },
+    {
+        "sentence": "To give you a sense of cell size, a typical human red blood cell is about eight millionths of a meter or eight micrometers (abbreviated as eight \u03bcm) in diameter.",
+        "triplets": [
+            "human red blood cell::has diameter::eight millionths of a meter or eight micrometers"
+        ]
+    },
+    {
+        "sentence": "A pin head is about two thousandths of a meter (two mm) in diameter.",
+        "triplets": [
+            "pin head::has diameter::two thousandths of a meter"
+        ]
+    },
+    {
+        "sentence": "That means about 250 red blood cells could fit on a pinhead.",
+        "triplets": [
+            "red blood cells::fit on::a pinhead"
+        ]
+    },
+    {
+        "sentence": "Most student microscopes are light microscopes (Figure 4.2a).",
+        "triplets": [
+            "Most student microscopes::are::light microscopes"
+        ]
+    },
+    {
+        "sentence": "Visible light passes and bends through the lens system to enable the user to see the specimen.",
+        "triplets": [
+            "Visible light::passes through::the lens system",
+            "Visible light::bends through::the lens system",
+            "the lens system::enables::the user to see the specimen"
+        ]
+    },
+    {
+        "sentence": "Light microscopes are advantageous for viewing living organisms, but since individual cells are generally transparent, their components are not distinguishable unless they are colored with special stains.",
+        "triplets": [
+            "Light microscopes::are advantageous for viewing::living organisms",
+            "Components of cells::are not distinguishable unless colored with::special stains"
+        ]
+    },
+    {
+        "sentence": "Staining, however, usually kills the cells.",
+        "triplets": [
+            "Staining::kills::cells"
+        ]
+    },
+    {
+        "sentence": "Light microscopes that undergraduates commonly use in the laboratory magnify up to approximately 400 times.",
+        "triplets": [
+            "Light microscopes::commonly used by::undergraduates in the laboratory",
+            "Light microscopes::magnify::up to approximately 400 times"
+        ]
+    },
+    {
+        "sentence": "Two parameters that are important in microscopy are magnification and resolving power.",
+        "triplets": [
+            "microscopy::has_parameter::magnification",
+            "microscopy::has_parameter::resolving power"
+        ]
+    },
+    {
+        "sentence": "Magnification is the process of enlarging an object in appearance.",
+        "triplets": [
+            "Magnification::is the process of::enlarging an object in appearance"
+        ]
+    },
+    {
+        "sentence": "Resolving power is the microscope's ability to distinguish two adjacent structures as separate: the higher the resolution, the better the image's clarity and detail.",
+        "triplets": [
+            "Resolving power::is the ability to distinguish::two adjacent structures as separate",
+            "resolution::is higher in::better image's clarity and detail"
+        ]
+    },
+    {
+        "sentence": "When one uses oil immersion lenses to study small objects, magnification usually increases to 1,000 times.",
+        "triplets": [
+            "oil immersion lenses::used to study::small objects",
+            "oil immersion lenses::increases magnification to::1"
+        ]
+    },
+    {
+        "sentence": "In order to gain a better understanding of cellular structure and function, scientists typically use electron microscopes.",
+        "triplets": [
+            "scientists::use::electron microscopes",
+            "scientists::gain understanding of::cellular structure and function"
+        ]
+    },
+    {
+        "sentence": "a) Most light microscopes in a college biology lab can magnify cells up to approximately 400 times and have a resolution of about 200 nanometers.",
+        "triplets": [
+            "Most light microscopes in a college biology lab::can magnify::cells up to approximately 400 times",
+            "Most light microscopes in a college biology lab::have::a resolution of about 200 nanometers"
+        ]
+    },
+    {
+        "sentence": "b) Electron microscopes provide a much higher magnification, 100,000x, and a have a resolution of 50 picometers.",
+        "triplets": [
+            "Electron microscopes::provide::a much higher magnification",
+            "Electron microscopes::have::a resolution of 50 picometers"
+        ]
+    },
+    {
+        "sentence": "use a beam of electrons instead of a beam of light.",
+        "triplets": [
+            "beam of electrons::can be used instead of::beam of light"
+        ]
+    },
+    {
+        "sentence": "Not only does this allow for higher magnification and, thus, more detail (Figure 4.3), it also provides higher resolving power.",
+        "triplets": [
+            "this::allow for::higher magnification and more detail",
+            "this::provide::higher resolving power"
+        ]
+    },
+    {
+        "sentence": "The method to prepare the specimen for viewing with an electron microscope kills the specimen.",
+        "triplets": [
+            "method::prepares::specimen",
+            "method::kills::specimen"
+        ]
+    },
+    {
+        "sentence": "Electrons have short wavelengths (shorter than photons) that move best in a vacuum, so we cannot view living cells with an electron microscope.",
+        "triplets": [
+            "Electrons::have::short wavelengths",
+            "Electrons::move best in::a vacuum",
+            "We::cannot view::living cells with an electron microscope"
+        ]
+    },
+    {
+        "sentence": "In a scanning electron microscope, a beam of electrons moves back and forth across a cell's surface, creating details of cell surface characteristics.",
+        "triplets": [
+            "scanning electron microscope::uses::beam of electrons",
+            "beam of electrons::moves across::cell's surface",
+            "movement of beam::creates::details of cell surface characteristics"
+        ]
+    },
+    {
+        "sentence": "In a transmission electron microscope, the electron beam penetrates the cell and provides details of a cell's internal structures.",
+        "triplets": [
+            "transmission electron microscope::has electron beam::penetrates cell",
+            "electron beam::provides details of::cell's internal structures"
+        ]
+    },
+    {
+        "sentence": "As you might imagine, electron microscopes are significantly more bulky and expensive than light microscopes.",
+        "triplets": [
+            "electron microscopes::are significantly more bulky and expensive than::light microscopes"
+        ]
+    },
+    {
+        "sentence": "a) These Salmonella bacteria appear as tiny purple dots when viewed with a light microscope.",
+        "triplets": [
+            "Salmonella bacteria::appear as::tiny purple dots"
+        ]
+    },
+    {
+        "sentence": "b) This scanning electron microscope micrograph shows Salmonella bacteria (in red) invading human cells (yellow).",
+        "triplets": [
+            "Salmonella bacteria::invade::human cells",
+            "scanning electron microscope micrograph::show::Salmonella bacteria invading human cells"
+        ]
+    },
+    {
+        "sentence": "Even though subfigure (b) shows a different Salmonella specimen than subfigure (a), you can still observe the comparative increase in magnification and detail.",
+        "triplets": [
+            "you::can observe::the comparative increase in magnification and detail"
+        ]
+    },
+    {
+        "sentence": "For another perspective on cell size, try the HowBig interactive at this site.",
+        "triplets": [
+            "HowBig interactive::provides perspective on::cell size"
+        ]
+    },
+    {
+        "sentence": "The microscopes we use today are far more complex than those that Dutch shopkeeper Antony van Leeuwenhoek, used in the 1600s.",
+        "triplets": [
+            "The microscopes::are more complex than::those that Dutch shopkeeper Antony van Leeuwenhoek used in the 1600s"
+        ]
+    },
+    {
+        "sentence": "Skilled in crafting lenses, van Leeuwenhoek observed the movements of single-celled organisms, which he collectively termed \"animalcules.\"",
+        "triplets": [
+            "van Leeuwenhoek::skilled in::crafting lenses",
+            "van Leeuwenhoek::observed::movements of single-celled organisms"
+        ]
+    },
+    {
+        "sentence": "In the 1665 publication Micrographia, experimental scientist Robert Hooke coined the term \"cell\" for the box-like structures he observed when viewing cork tissue through a lens.",
+        "triplets": [
+            "Robert Hooke::coined::term",
+            "Robert Hooke::observed::box-like structures"
+        ]
+    },
+    {
+        "sentence": "In the 1670s, van Leeuwenhoek discovered bacteria and protozoa.",
+        "triplets": [
+            "van Leeuwenhoek::discovered::bacteria",
+            "van Leeuwenhoek::discovered::protozoa"
+        ]
+    },
+    {
+        "sentence": "Later advances in lenses, microscope construction, and staining techniques enabled other scientists to see some components inside cells.",
+        "triplets": [
+            "Later advances in lenses::enabled scientists to see::some components inside cells",
+            "Microscope construction::enabled scientists to see::some components inside cells",
+            "Staining techniques::enabled scientists to see::some components inside cells"
+        ]
+    },
+    {
+        "sentence": "By the late 1830s, botanist Matthias Schleiden and zoologist Theodor Schwann were studying tissues and proposed the unified cell theory, which states that one or more cells comprise all living things, the cell is the basic unit of life, and new cells arise from existing cells.",
+        "triplets": [
+            "Matthias Schleiden::studied::tissues",
+            "Theodor Schwann::studied::tissues",
+            "Matthias Schleiden::proposed::unified cell theory",
+            "Theodor Schwann::proposed::unified cell theory",
+            "unified cell theory::states::one or more cells comprise all living things",
+            "unified cell theory::states::the cell is the basic unit of life",
+            "unified cell theory::states::new cells arise from existing cells"
+        ]
+    },
+    {
+        "sentence": "Rudolf Virchow later made important contributions to this theory.",
+        "triplets": [
+            "Rudolf Virchow::made contributions to::this theory"
+        ]
+    },
+    {
+        "sentence": "Have you ever heard of a medical test called a Pap smear (Figure 4.4)?",
+        "triplets": [
+            "Pap smear::is a medical test::exists"
+        ]
+    },
+    {
+        "sentence": "In this test, a doctor takes a small sample of cells from the patient's uterine cervix and sends it to a medical lab where a cytotechnologist stains the cells and examines them for any changes that could indicate cervical cancer or a microbial infection.",
+        "triplets": [
+            "doctor::takes::small sample of cells from patient's uterine cervix",
+            "doctor::sends::sample to medical lab",
+            "cytotechnologist::stains::cells",
+            "cytotechnologist::examines::stained cells for changes",
+            "changes in cells::indicate::cervical cancer or microbial infection"
+        ]
+    },
+    {
+        "sentence": "Cytotechnologists (cyto- = \"cell\") are professionals who study cells via microscopic examinations and other laboratory tests.",
+        "triplets": [
+            "Cytotechnologists::study::cells",
+            "Cytotechnologists::are::professionals"
+        ]
+    },
+    {
+        "sentence": "They are trained to determine which cellular changes are within normal limits and which are abnormal.",
+        "triplets": [
+            "They::are trained to determine::which cellular changes are within normal limits",
+            "They::are trained to determine::which cellular changes are abnormal"
+        ]
+    },
+    {
+        "sentence": "Their focus is not limited to cervical cells.",
+        "triplets": [
+            "Their::focus::not limited to cervical cells"
+        ]
+    },
+    {
+        "sentence": "They study cellular specimens that come from all organs.",
+        "triplets": [
+            "They::study::cellular specimens",
+            "cellular specimens::come from::all organs"
+        ]
+    },
+    {
+        "sentence": "When they notice abnormalities, they consult a pathologist, a medical doctor who interprets and diagnoses changes that disease in body tissue and fluids cause.",
+        "triplets": [
+            "they::consult::pathologist",
+            "pathologist::is a::medical doctor",
+            "pathologist::interprets and diagnoses::changes that disease in body tissue and fluids cause"
+        ]
+    },
+    {
+        "sentence": "Cytotechnologists play a vital role in saving people's lives.",
+        "triplets": [
+            "Cytotechnologists::play a vital role in::saving people's lives",
+            "Cytotechnologists::save::people's lives"
+        ]
+    },
+    {
+        "sentence": "When doctors discover abnormalities early, a patient's treatment can begin sooner, which usually increases the chances of a successful outcome.",
+        "triplets": [
+            "doctors::discover::abnormalities early",
+            "early discovery of abnormalities::allow for::earlier treatment",
+            "earlier treatment::increase chances of::successful outcome"
+        ]
+    },
+    {
+        "sentence": "These uterine cervix cells, viewed through a light microscope, are from a Pap smear.",
+        "triplets": [
+            "uterine cervix cells::are from::Pap smear",
+            "cells::are::uterine cervix cells",
+            "cells::were viewed through::light microscope"
+        ]
+    },
+    {
+        "sentence": "Normal cells are on the left.",
+        "triplets": [
+            "Normal cells::are on::the left"
+        ]
+    },
+    {
+        "sentence": "The cells on the right are infected with human papillomavirus (HPV).",
+        "triplets": [
+            "cells on the right::are infected with::human papillomavirus"
+        ]
+    },
+    {
+        "sentence": "Notice that the infected cells are larger.",
+        "triplets": []
+    },
+    {
+        "sentence": "Also, two of these cells each have two nuclei instead of one, the normal number.",
+        "triplets": [
+            "cells::have nuclei::two",
+            "cells::have nuclei::one",
+            "cells::have number::normal"
+        ]
+    }
+]

output/2023-08-28_triplets_Traditional.json

@@ -0,0 +1,532 @@
+[
+    {
+        "sentence": "A cell is the smallest unit of a living thing.",
+        "triplets": [
+            "cell::is the smallest unit of::living thing"
+        ]
+    },
+    {
+        "sentence": "Whether comprised of one cell (like bacteria) or many cells (like a human), we call it an organism.",
+        "triplets": [
+            "one cell::comprised of::bacteria",
+            "many cells::comprised of::human"
+        ]
+    },
+    {
+        "sentence": "Thus, cells are the basic building blocks of all organisms.",
+        "triplets": [
+            "cells::are the basic building blocks of::organisms"
+        ]
+    },
+    {
+        "sentence": "Several cells of one kind that interconnect with each other and perform a shared function form tissues.",
+        "triplets": [
+            "cells::form::tissues"
+        ]
+    },
+    {
+        "sentence": "These tissues combine to form an organ (your stomach, heart, or brain), and several organs comprise an organ system (such as the digestive system, circulatory system, or nervous system).",
+        "triplets": [
+            "tissues::combine to form::organ",
+            "organ::comprise::organ system",
+            "organ system::comprise::N",
+            "organ system::such as::digestive system",
+            "organ system::such as::circulatory system",
+            "organ system::such as::nervous system"
+        ]
+    },
+    {
+        "sentence": "Several systems that function together form an organism (like a human being).",
+        "triplets": [
+            "systems::form::organism",
+            "systems::function together::organism",
+            "systems::form::human being",
+            "systems::function together::human being",
+            "organism::like::human being"
+        ]
+    },
+    {
+        "sentence": "Here, we will examine the structure and function of cells.",
+        "triplets": [
+            "structure::is related to::cells",
+            "function::is related to::cells"
+        ]
+    },
+    {
+        "sentence": "There are many types of cells, which scientists group into one of two broad categories: prokaryotic and eukaryotic.",
+        "triplets": [
+            "types of cells::grouped into::categories",
+            "scientists::group into::categories",
+            "categories::have::prokaryotic",
+            "categories::have::eukaryotic"
+        ]
+    },
+    {
+        "sentence": "For example, we classify both animal and plant cells as eukaryotic cells; whereas, we classify bacterial cells as prokaryotic.",
+        "triplets": [
+            "animal cells::classified as::eukaryotic cells",
+            "plant cells::classified as::eukaryotic cells",
+            "bacterial cells::classified as::prokaryotic"
+        ]
+    },
+    {
+        "sentence": "Before discussing the criteria for determining whether a cell is prokaryotic or eukaryotic, we will first examine how biologists study cells.",
+        "triplets": [
+            "criteria::determines::cell",
+            "cell::is::prokaryotic",
+            "cell::is::eukaryotic",
+            "biologists::study::cells"
+        ]
+    },
+    {
+        "sentence": "Cells vary in size.",
+        "triplets": [
+            "Cells::vary in::size"
+        ]
+    },
+    {
+        "sentence": "With few exceptions, we cannot see individual cells with the naked eye, so scientists use microscopes (micro- = \"small\"; -scope = \"to look at\") to study them.",
+        "triplets": []
+    },
+    {
+        "sentence": "A microscope is an instrument that magnifies an object.",
+        "triplets": [
+            "microscope::is an instrument::object",
+            "microscope::magnifies::object"
+        ]
+    },
+    {
+        "sentence": "We photograph most cells with a microscope, so we can call these images micrographs.",
+        "triplets": [
+            "cells::photograph::microscope",
+            "cells::call::images",
+            "images::be::micrographs"
+        ]
+    },
+    {
+        "sentence": "The optics of a microscope's lenses change the image orientation that the user sees.",
+        "triplets": [
+            "optics::change::image orientation",
+            "microscope's lenses::change::image orientation",
+            "image orientation::seen by::user"
+        ]
+    },
+    {
+        "sentence": "A specimen that is right-side up and facing right on the microscope slide will appear upside-down and facing left when one views through a microscope, and vice versa.",
+        "triplets": [
+            "Specimen::is placed on::Microscope slide",
+            "Specimen::appears as::Upside-down",
+            "Specimen::appears as::Facing left",
+            "Microscope slide::is viewed through::Microscope",
+            "Microscope::causes::Specimen to appear upside-down",
+            "Microscope::causes::Specimen to appear facing left"
+        ]
+    },
+    {
+        "sentence": "Similarly, if one moves the slide left while looking through the microscope, it will appear to move right, and if one moves it down, it will seem to move up.",
+        "triplets": []
+    },
+    {
+        "sentence": "This occurs because microscopes use two sets of lenses to magnify the image.",
+        "triplets": [
+            "microscopes::use::two sets of lenses",
+            "microscopes::magnify::image",
+            "sets of lenses::magnify::image"
+        ]
+    },
+    {
+        "sentence": "Because of the manner by which light travels through the lenses, this two lens system produces an inverted image (binocular, or dissecting microscopes, work in a similar manner, but include an additional magnification system that makes the final image appear to be upright).",
+        "triplets": [
+            "lenses::travels through::light",
+            "two lens system::produces::inverted image",
+            "binocular::work in a similar manner as::dissecting microscopes",
+            "dissecting microscopes::include::additional magnification system",
+            "additional magnification system::makes the final image appear to be::upright"
+        ]
+    },
+    {
+        "sentence": "To give you a sense of cell size, a typical human red blood cell is about eight millionths of a meter or eight micrometers (abbreviated as eight \u03bcm) in diameter.",
+        "triplets": [
+            "cell size::is about::eight millionths of a meter",
+            "cell size::is about::eight micrometers",
+            "human red blood cell::is::eight millionths of a meter",
+            "human red blood cell::is::eight micrometers",
+            "eight millionths of a meter::is in diameter::human red blood cell",
+            "eight micrometers::is in diameter::human red blood cell"
+        ]
+    },
+    {
+        "sentence": "A pin head is about two thousandths of a meter (two mm) in diameter.",
+        "triplets": [
+            "pin head::has diameter::two thousandths of a meter",
+            "two thousandths of a meter::is measured in::mm"
+        ]
+    },
+    {
+        "sentence": "That means about 250 red blood cells could fit on a pinhead.",
+        "triplets": [
+            "250 red blood cells::could fit on::pinhead"
+        ]
+    },
+    {
+        "sentence": "Most student microscopes are light microscopes (Figure 4.2a).",
+        "triplets": [
+            "student microscopes::are::light microscopes",
+            "student microscopes::are::N",
+            "student microscopes::are::Figure 4.2a",
+            "light microscopes::are::student microscopes",
+            "light microscopes::are::N",
+            "light microscopes::are::Figure 4.2a",
+            "Figure 4.2a::are::student microscopes",
+            "Figure 4.2a::are::light microscopes",
+            "Figure 4.2a::are::N"
+        ]
+    },
+    {
+        "sentence": "Visible light passes and bends through the lens system to enable the user to see the specimen.",
+        "triplets": [
+            "Visible light::passes through::lens system",
+            "Visible light::bends through::lens system",
+            "lens system::enables::user",
+            "user::sees::specimen"
+        ]
+    },
+    {
+        "sentence": "Light microscopes are advantageous for viewing living organisms, but since individual cells are generally transparent, their components are not distinguishable unless they are colored with special stains.",
+        "triplets": [
+            "Light microscopes::are advantageous for::viewing living organisms",
+            "individual cells::are not distinguishable unless colored with::special stains",
+            "components::are colored with::special stains"
+        ]
+    },
+    {
+        "sentence": "Staining, however, usually kills the cells.",
+        "triplets": [
+            "Staining::kills::cells"
+        ]
+    },
+    {
+        "sentence": "Light microscopes that undergraduates commonly use in the laboratory magnify up to approximately 400 times.",
+        "triplets": [
+            "Light microscopes::commonly use in::laboratory",
+            "undergraduates::commonly use in::laboratory",
+            "Light microscopes::magnify up to::approximately 400 times"
+        ]
+    },
+    {
+        "sentence": "Two parameters that are important in microscopy are magnification and resolving power.",
+        "triplets": [
+            "parameters::are important in::microscopy",
+            "microscopy::are important in::parameters",
+            "parameters::are important in::magnification",
+            "magnification::are important in::parameters",
+            "parameters::are important in::resolving power",
+            "resolving power::are important in::parameters"
+        ]
+    },
+    {
+        "sentence": "Magnification is the process of enlarging an object in appearance.",
+        "triplets": [
+            "Magnification::is the process of::enlarging an object in appearance",
+            "process::enlarging::an object in appearance",
+            "object::in::appearance"
+        ]
+    },
+    {
+        "sentence": "Resolving power is the microscope's ability to distinguish two adjacent structures as separate: the higher the resolution, the better the image's clarity and detail.",
+        "triplets": [
+            "Resolving power::is the ability of::microscope's ability",
+            "Resolving power::distinguish::two adjacent structures",
+            "resolution::is higher with::better image's clarity",
+            "resolution::is higher with::better detail",
+            "image's clarity::is better with::higher resolution",
+            "detail::is better with::higher resolution"
+        ]
+    },
+    {
+        "sentence": "When one uses oil immersion lenses to study small objects, magnification usually increases to 1,000 times.",
+        "triplets": [
+            "oil immersion lenses::used to study::small objects",
+            "oil immersion lenses::increases magnification to::1",
+            "small objects::studied using::oil immersion lenses",
+            "magnification::increases to::1"
+        ]
+    },
+    {
+        "sentence": "In order to gain a better understanding of cellular structure and function, scientists typically use electron microscopes.",
+        "triplets": [
+            "cellular structure::is used for::understanding",
+            "cellular structure::is used for::function",
+            "scientists::use::electron microscopes"
+        ]
+    },
+    {
+        "sentence": "a) Most light microscopes in a college biology lab can magnify cells up to approximately 400 times and have a resolution of about 200 nanometers.",
+        "triplets": [
+            "light microscopes::magnify::cells",
+            "light microscopes::have::resolution",
+            "light microscopes::have::nanometers",
+            "college biology lab::contain::light microscopes",
+            "cells::magnified by::light microscopes",
+            "resolution::measured in::nanometers"
+        ]
+    },
+    {
+        "sentence": "b) Electron microscopes provide a much higher magnification, 100,000x, and a have a resolution of 50 picometers.",
+        "triplets": [
+            "Electron microscopes::provide::magnification",
+            "Electron microscopes::have::resolution",
+            "resolution::is of::picometers"
+        ]
+    },
+    {
+        "sentence": "use a beam of electrons instead of a beam of light.",
+        "triplets": [
+            "beam of electrons::instead of::beam of light"
+        ]
+    },
+    {
+        "sentence": "Not only does this allow for higher magnification and, thus, more detail (Figure 4.3), it also provides higher resolving power.",
+        "triplets": [
+            "higher magnification::allows for::more detail",
+            "higher magnification::provides::higher resolving power"
+        ]
+    },
+    {
+        "sentence": "The method to prepare the specimen for viewing with an electron microscope kills the specimen.",
+        "triplets": [
+            "method::prepares::specimen",
+            "method::kills::specimen",
+            "specimen::is prepared for::viewing",
+            "specimen::is viewed with::electron microscope"
+        ]
+    },
+    {
+        "sentence": "Electrons have short wavelengths (shorter than photons) that move best in a vacuum, so we cannot view living cells with an electron microscope.",
+        "triplets": [
+            "Electrons::have::short wavelengths",
+            "Electrons::move best in::vacuum",
+            "we::cannot view::living cells",
+            "we::cannot view::electron microscope",
+            "short wavelengths::are shorter than::photons",
+            "vacuum::is best for::electrons",
+            "living cells::cannot be viewed with::electron microscope"
+        ]
+    },
+    {
+        "sentence": "In a scanning electron microscope, a beam of electrons moves back and forth across a cell's surface, creating details of cell surface characteristics.",
+        "triplets": [
+            "scanning electron microscope::moves across::beam of electrons",
+            "beam of electrons::creates::details",
+            "beam of electrons::moves across::cell's surface",
+            "details::of::cell surface characteristics"
+        ]
+    },
+    {
+        "sentence": "In a transmission electron microscope, the electron beam penetrates the cell and provides details of a cell's internal structures.",
+        "triplets": [
+            "transmission electron microscope::penetrates::cell",
+            "electron beam::provides details of::cell's internal structures"
+        ]
+    },
+    {
+        "sentence": "As you might imagine, electron microscopes are significantly more bulky and expensive than light microscopes.",
+        "triplets": [
+            "electron microscopes::are significantly more bulky and expensive than::light microscopes"
+        ]
+    },
+    {
+        "sentence": "a) These Salmonella bacteria appear as tiny purple dots when viewed with a light microscope.",
+        "triplets": [
+            "Salmonella bacteria::appear as::tiny purple dots",
+            "tiny purple dots::viewed with::light microscope"
+        ]
+    },
+    {
+        "sentence": "b) This scanning electron microscope micrograph shows Salmonella bacteria (in red) invading human cells (yellow).",
+        "triplets": [
+            "scanning electron microscope::shows::Salmonella bacteria",
+            "scanning electron microscope::shows::human cells",
+            "micrograph::shows::Salmonella bacteria",
+            "micrograph::shows::human cells",
+            "Salmonella bacteria::invading::human cells"
+        ]
+    },
+    {
+        "sentence": "Even though subfigure (b) shows a different Salmonella specimen than subfigure (a), you can still observe the comparative increase in magnification and detail.",
+        "triplets": []
+    },
+    {
+        "sentence": "For another perspective on cell size, try the HowBig interactive at this site.",
+        "triplets": [
+            "perspective::on::cell size",
+            "perspective::at::site",
+            "HowBig interactive::at::site"
+        ]
+    },
+    {
+        "sentence": "The microscopes we use today are far more complex than those that Dutch shopkeeper Antony van Leeuwenhoek, used in the 1600s.",
+        "triplets": [
+            "microscopes::are more complex than::Dutch shopkeeper Antony van Leeuwenhoek"
+        ]
+    },
+    {
+        "sentence": "Skilled in crafting lenses, van Leeuwenhoek observed the movements of single-celled organisms, which he collectively termed \"animalcules.\"",
+        "triplets": [
+            "van Leeuwenhoek::observed::movements",
+            "van Leeuwenhoek::observed::single-celled organisms",
+            "van Leeuwenhoek::termed::animalcules",
+            "lenses::crafted by::van Leeuwenhoek"
+        ]
+    },
+    {
+        "sentence": "In the 1665 publication Micrographia, experimental scientist Robert Hooke coined the term \"cell\" for the box-like structures he observed when viewing cork tissue through a lens.",
+        "triplets": [
+            "Micrographia::publication::1665",
+            "Robert Hooke::coined::term",
+            "Robert Hooke::observed::box-like structures",
+            "Robert Hooke::viewing::cork tissue",
+            "Robert Hooke::viewing::lens",
+            "cork tissue::viewed through::lens"
+        ]
+    },
+    {
+        "sentence": "In the 1670s, van Leeuwenhoek discovered bacteria and protozoa.",
+        "triplets": [
+            "van Leeuwenhoek::discovered::bacteria",
+            "van Leeuwenhoek::discovered::protozoa"
+        ]
+    },
+    {
+        "sentence": "Later advances in lenses, microscope construction, and staining techniques enabled other scientists to see some components inside cells.",
+        "triplets": [
+            "advances::in::lenses",
+            "advances::in::microscope construction",
+            "advances::in::staining techniques",
+            "lenses::enabled::scientists",
+            "microscope construction::enabled::scientists",
+            "staining techniques::enabled::scientists",
+            "scientists::see::components",
+            "components::inside::cells"
+        ]
+    },
+    {
+        "sentence": "By the late 1830s, botanist Matthias Schleiden and zoologist Theodor Schwann were studying tissues and proposed the unified cell theory, which states that one or more cells comprise all living things, the cell is the basic unit of life, and new cells arise from existing cells.",
+        "triplets": [
+            "botanist Matthias Schleiden::studied::tissues",
+            "zoologist Theodor Schwann::studied::tissues",
+            "botanist Matthias Schleiden::proposed::unified cell theory",
+            "zoologist Theodor Schwann::proposed::unified cell theory",
+            "unified cell theory::states that::one or more cells comprise all living things",
+            "unified cell theory::states that::the cell is the basic unit of life",
+            "unified cell theory::states that::new cells arise from existing cells"
+        ]
+    },
+    {
+        "sentence": "Rudolf Virchow later made important contributions to this theory.",
+        "triplets": [
+            "Rudolf Virchow::made contributions to::theory"
+        ]
+    },
+    {
+        "sentence": "Have you ever heard of a medical test called a Pap smear (Figure 4.4)?",
+        "triplets": [
+            "medical test::called::Pap smear"
+        ]
+    },
+    {
+        "sentence": "In this test, a doctor takes a small sample of cells from the patient's uterine cervix and sends it to a medical lab where a cytotechnologist stains the cells and examines them for any changes that could indicate cervical cancer or a microbial infection.",
+        "triplets": [
+            "doctor::takes::small sample",
+            "doctor::sends::cells",
+            "cells::stains::cytotechnologist",
+            "cells::examines::changes",
+            "changes::indicate::cervical cancer",
+            "changes::indicate::microbial infection"
+        ]
+    },
+    {
+        "sentence": "Cytotechnologists (cyto- = \"cell\") are professionals who study cells via microscopic examinations and other laboratory tests.",
+        "triplets": [
+            "Cytotechnologists::are professionals::N",
+            "Cytotechnologists::study::cells",
+            "professionals::study::cells",
+            "cells::via::microscopic examinations",
+            "cells::via::other laboratory tests"
+        ]
+    },
+    {
+        "sentence": "They are trained to determine which cellular changes are within normal limits and which are abnormal.",
+        "triplets": []
+    },
+    {
+        "sentence": "Their focus is not limited to cervical cells.",
+        "triplets": [
+            "Their::has_focus::focus",
+            "focus::has_focus_on::cervical cells"
+        ]
+    },
+    {
+        "sentence": "They study cellular specimens that come from all organs.",
+        "triplets": [
+            "cellular specimens::come from::organs"
+        ]
+    },
+    {
+        "sentence": "When they notice abnormalities, they consult a pathologist, a medical doctor who interprets and diagnoses changes that disease in body tissue and fluids cause.",
+        "triplets": [
+            "abnormalities::notice::they",
+            "they::consult::pathologist",
+            "pathologist::interprets and diagnoses::changes",
+            "pathologist::interprets and diagnoses::disease",
+            "changes::cause::disease",
+            "disease::cause::body tissue",
+            "disease::cause::fluids"
+        ]
+    },
+    {
+        "sentence": "Cytotechnologists play a vital role in saving people's lives.",
+        "triplets": [
+            "Cytotechnologists::play a vital role in::people's lives"
+        ]
+    },
+    {
+        "sentence": "When doctors discover abnormalities early, a patient's treatment can begin sooner, which usually increases the chances of a successful outcome.",
+        "triplets": [
+            "doctors::discover::abnormalities",
+            "doctors::begin::treatment",
+            "treatment::increase::chances",
+            "treatment::lead to::successful outcome"
+        ]
+    },
+    {
+        "sentence": "These uterine cervix cells, viewed through a light microscope, are from a Pap smear.",
+        "triplets": [
+            "uterine cervix cells::viewed through::light microscope",
+            "uterine cervix cells::from::Pap smear"
+        ]
+    },
+    {
+        "sentence": "Normal cells are on the left.",
+        "triplets": [
+            "Normal cells::are on::left"
+        ]
+    },
+    {
+        "sentence": "The cells on the right are infected with human papillomavirus (HPV).",
+        "triplets": [
+            "cells::infected with::human papillomavirus"
+        ]
+    },
+    {
+        "sentence": "Notice that the infected cells are larger.",
+        "triplets": []
+    },
+    {
+        "sentence": "Also, two of these cells each have two nuclei instead of one, the normal number.",
+        "triplets": [
+            "cells::have::two nuclei",
+            "cells::have::normal number"
+        ]
+    }
+]

pipeline.py

@@ -0,0 +1,87 @@
+import time
+from my_lib.chatbot import ChatGPT
+import re
+
+
+class BasePipeline:
+    
+    chat = ChatGPT(model="gpt-3.5-turbo")
+
+    def add_task(self, task, **inputs):
+        args = task["args"]
+        args_supplied = {arg: arg in inputs for arg in args}
+        
+        if all(args_supplied.values()):
+
+            formatted_task = f"[TASK]\n{task['description']}"
+
+            if len(args):
+                formatted_task += f"\n\n[INPUT]\n" + '\n'.join([f'{arg}: {inputs[arg]}' for arg in args])
+
+            self.pipe.append(formatted_task)
+        else:
+            raise ValueError(
+                f"Missing required argument(s): {[arg for arg in args if not args_supplied[arg]]}"
+            )
+    
+
+    def compile_tasks(self, inputs_by_idx):
+        self.pipe = []
+
+        for idx, task in enumerate(self.prompt):
+            self.add_task(task, **inputs_by_idx.get(idx, {}))
+    
+    
+    def forward(self):
+        self.history = []
+        
+        for task in self.pipe:
+            output = self.chat(task)
+            time.sleep(0.2)
+
+        self.history.append(self.chat.history)
+        self.chat.clear_history()
+        
+        return self.postprocess(output)
+    
+
+    def postprocess(self):
+        raise NotImplementedError
+
+
+class Text2KG(BasePipeline):
+
+    
+    def __init__(self, recipe: dict):
+
+        self.name = recipe["name"]
+        self.prompt = recipe["prompt"]
+
+        self.chat = ChatGPT(
+            model="gpt-3.5-turbo",
+            init=(
+                "You are a sentence parsing agent helping to construct a knowledge graph."
+            ),
+            temperature=0.3
+        )
+
+    
+    def __call__(self, text):
+        self.compile_tasks(
+            inputs_by_idx={0: {"text": text}}
+        )
+
+        return self.forward()
+    
+
+    def __repr__(self):
+        return f"Text2KG(recipe={self.name})"
+
+    
+    def postprocess(self, output: str):
+        
+        word_pattern = r"'?\w+(?:[ |.'-]\w+)*'?"
+        triplet_pattern = f'({word_pattern}::{word_pattern}::{word_pattern})'
+        processed = re.findall(triplet_pattern, re.sub(r'[<>]', '', output))
+        
+        return processed

recipes.json

@@ -0,0 +1,37 @@
+[
+    {
+        "name": "LogicBased",
+        "prompt": [
+            {
+                "description": "Given the text, extract a list of the premises embedded within it. Focus on identifying declarative sentences that convey factual information.",
+                "args": ["text"]
+            },
+            {
+                "description": "Convert each premise into a knowledge graph triplet consisting of a subject, predicate, and object. The subject should be the main entity mentioned, the predicate should represent the relationship or attribute, and the object should be another entity or value associated with the subject. Return a list of these knowledge graph triplets, where each triplet has the form: <subject>::<relation>::<object>",
+                "args": []
+            }
+        ]
+    },
+    {
+        "name": "Traditional",
+        "prompt": [
+            {
+                "description": "Extract the entities from the text. Focus on descriptive noun chunks and noun phrases.",
+                "args": ["text"]
+            },
+            {
+                "description": "Label all pairs of entities with the appropriate relation between them (or N/A if none). The result will be a list of triplets of the form: <subject>::<relation>::<object>",
+                "args": []
+            }
+        ]
+    },
+    {
+        "name": "Direct",
+        "prompt": [
+            {
+                "description": "Extract a list of knowledge graph triplets from the text. Triplets should have the form: <subject>::<relation>::<object>.",
+                "args": ["text"]
+            }
+        ]
+    }
+]

text2kg.py

@@ -0,0 +1,69 @@
+from argparse import ArgumentParser
+from datetime import date
+import json
+import os
+
+from nltk.tokenize import sent_tokenize
+from tqdm import tqdm
+
+from pipeline import Text2KG
+
+
+def parse_args():
+    parser = ArgumentParser()
+    parser.add_argument("infile", type=str)
+    parser.add_argument("--output", type=str, default="./output")
+    parser.add_argument("--cookbook", type=str, default="./recipes.json",
+                        help="path to cookbook")
+    parser.add_argument("--recipe", type=str, default=None,
+                        help="name of recipe to use"),
+    parser.add_argument("--thoughts", action="store_true")
+    
+    return parser.parse_args()
+
+
+def save(name, item, args):
+
+    os.makedirs(args.output, exist_ok=True)
+
+    today = date.today()
+    filename = f"{today}_{name}_{args.recipe}.json"
+    filepath = os.path.join(args.output, filename)
+
+    with open(filepath, 'w') as f:
+        json.dump(item, f)
+
+
+def main(args):
+    with open(args.cookbook) as f:
+        cookbook = json.load(f)
+    
+    recipe = None
+    for item in cookbook:
+        if item["name"] == args.recipe:
+            recipe = item
+    if recipe is None:
+        raise ValueError(f"Recipe '{args.recipe}' does not exist in cookbook.")
+        
+    pipe = Text2KG(recipe)
+
+    with open(args.infile) as f:
+        text = f.read()
+
+    sentences = sent_tokenize(text.replace('\n', ' '))
+    
+    triplets = [pipe(s) for s in tqdm(sentences)]
+
+    output = [{"sentence": s, "triplets": t} for s, t in zip(sentences, triplets)]
+
+    save("triplets", output, args)
+
+    if args.thoughts:
+        save("thoughts", pipe.history, args)
+    
+    return output
+
+
+if __name__ == "__main__":
+    args = parse_args()
+    main(args)