{ "paper_id": "2021", "header": { "generated_with": "S2ORC 1.0.0", "date_generated": "2023-01-19T10:46:46.418073Z" }, "title": "Finitestate script normalization and processing utilities: The Nisaba Brahmic library", "authors": [ { "first": "Cibu", "middle": [], "last": "Johny", "suffix": "", "affiliation": { "laboratory": "United Kingdom and \u2021 United States", "institution": "", "location": {} }, "email": "" }, { "first": "Lawrence", "middle": [], "last": "Wolfsonkin", "suffix": "", "affiliation": { "laboratory": "United Kingdom and \u2021 United States", "institution": "", "location": {} }, "email": "wolfsonkin@google.com" }, { "first": "Alexander", "middle": [], "last": "Gutkin", "suffix": "", "affiliation": { "laboratory": "United Kingdom and \u2021 United States", "institution": "", "location": {} }, "email": "agutkin@google.com" }, { "first": "Brian", "middle": [], "last": "Roark", "suffix": "", "affiliation": { "laboratory": "United Kingdom and \u2021 United States", "institution": "", "location": {} }, "email": "roark@google.com" }, { "first": "Google", "middle": [], "last": "Research", "suffix": "", "affiliation": { "laboratory": "United Kingdom and \u2021 United States", "institution": "", "location": {} }, "email": "" } ], "year": "", "venue": null, "identifiers": {}, "abstract": "This paper presents an opensource library for efficient lowlevel processing of ten ma jor South Asian Brahmic scripts. The library provides a flexible and extensible framework for supporting crucial operations on Brahmic scripts, such as NFC, visual normalization, reversible transliteration, and validity checks, implemented in Python within a finitestate transducer formalism. We survey some com mon Brahmic script issues that may adversely affect the performance of downstream NLP tasks, and provide the rationale for finitestate design and system implementation details.", "pdf_parse": { "paper_id": "2021", "_pdf_hash": "", "abstract": [ { "text": "This paper presents an opensource library for efficient lowlevel processing of ten ma jor South Asian Brahmic scripts. The library provides a flexible and extensible framework for supporting crucial operations on Brahmic scripts, such as NFC, visual normalization, reversible transliteration, and validity checks, implemented in Python within a finitestate transducer formalism. We survey some com mon Brahmic script issues that may adversely affect the performance of downstream NLP tasks, and provide the rationale for finitestate design and system implementation details.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Abstract", "sec_num": null } ], "body_text": [ { "text": "The Unicode Standard separates the representation of text from its specific graphical rendering: text is encoded as a sequence of characters, which, at presentation time are then collectively rendered into the appropriate sequence of glyphs for display. This can occasionally result in manytoone map pings, where several distinctlyencoded strings can result in identical display. For example, Latin script letters with diacritics such as \"\u00e9\" can gener ally be encoded as either: (a) a pair of the base let ter (e.g., \"e\" which is U+0065 from Unicode's Ba sic Latin block, corresponding to ASCII) and a dia critic (in this case U+0301 from the Combining Dia critical Marks block)\u037e or (b) a single character that represents the grapheme directly (U+00E9 from the Latin1 Supplement Unicode block). Both encod ings yield visually identical text, hence text is of ten normalized to a conventionalized normal form, such as the wellknown Normalization Form C (NFC), so that visually identical words are mapped to a conventionalized representative of their equiv alence class for downstream processing. Critically, NFC normalization falls far short of a complete handling of such manytoone phenomena in Uni code.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Introduction", "sec_num": "1" }, { "text": "In addition to such normalization issues, some scripts also have wellformedness constraints, i.e., not all strings of Unicode characters from a single script correspond to a valid (i.e., legible) grapheme sequence in the script. Such constraints do not ap ply in the basic Latin alphabet, where any permuta tion of letters can be rendered as a valid string (e.g., for use as an acronym). The Brahmic family of scripts, however, including the Devanagari script used to write Hindi, Marathi and many other South Asian languages, do have such constraints. These scripts are alphasyllabaries, meaning that they are structured around orthographic syllables (aks \u0323ara) as the basic unit. 1 One or more Unicode characters combine when rendering one of thousands of leg ible aks \u0323ara, but many combinations do not corre spond to any aks \u0323ara. Given a token in these scripts, one may want to (a) normalize it to a canonical form\u037e and (b) check whether it is a wellformed sequence of aks \u0323ara.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Introduction", "sec_num": "1" }, { "text": "Brahmic scripts are heavily used across South Asia and have official status in India, Bangladesh, Nepal, Sri Lanka and beyond (Cardona and Jain, 2007\u037e Steever, 2019) . Despite evident progress in localization standards (Unicode Consortium, 2019) and improvements in associated technolo gies such as input methods (Hinkle et al., 2013) and character recognition (Pal et al., 2012) , Brahmic script processing still poses important challenges due to the inherent differences between these writ ing systems and those which historically have been more dominant in information technology (Sinha, 2009\u037e Bhattacharyya et al., 2019 .", "cite_spans": [ { "start": 126, "end": 138, "text": "(Cardona and", "ref_id": "BIBREF6" }, { "start": 139, "end": 165, "text": "Jain, 2007\u037e Steever, 2019)", "ref_id": null }, { "start": 228, "end": 245, "text": "Consortium, 2019)", "ref_id": "BIBREF48" }, { "start": 313, "end": 334, "text": "(Hinkle et al., 2013)", "ref_id": "BIBREF19" }, { "start": 361, "end": 379, "text": "(Pal et al., 2012)", "ref_id": "BIBREF35" }, { "start": 583, "end": 623, "text": "(Sinha, 2009\u037e Bhattacharyya et al., 2019", "ref_id": null } ], "ref_spans": [], "eq_spans": [], "section": "Introduction", "sec_num": "1" }, { "text": "In this paper, we present Nisaba, an opensource software library, 2 which provides processing utili ties for ten major Brahmic scripts of South Asia: Bengali, Devanagari, Gujarati, Gurmukhi, Kan nada, Malayalam, Oriya (Odia), Sinhala, Tamil, and Telugu. In addition to string normaliza tion and wellformedness processing, the library also includes utilities for the deterministic and re versible romanization of these scripts, i.e., translit eration from each script to and from the Latin script (Wellisch, 1978) . While the resulting roman izations are standardized in a way that may or may not correspond to how native speakers tend to ro manize the text in informal communication (see, e.g., Roark et al., 2020) , such a default romaniza tion can permit easy inspection of an approximate version of the linguistic strings for those who read the Latin script but not the specific Brahmic script being examined.", "cite_spans": [ { "start": 496, "end": 512, "text": "(Wellisch, 1978)", "ref_id": "BIBREF49" }, { "start": 695, "end": 714, "text": "Roark et al., 2020)", "ref_id": "BIBREF39" } ], "ref_spans": [], "eq_spans": [], "section": "Introduction", "sec_num": "1" }, { "text": "As a whole, the library provides important utili ties for language processing applications of South Asian languages using Brahmic scripts. The de sign is based on the observation that, while there are considerable superficial differences between these scripts, they follow the same encoding model in Unicode, and maintain a very similar char acter repertoire having evolved from the same source -the Br\u0101hm\u012b script (Salomon, 1996\u037e Fe dorova, 2012 . This observation lends itself to the scriptagnostic design (outlined in \u00a74) that, unlike other approaches reviewed in \u00a72, is based on the weighted finite state transducer (WFST) formal ism (Mohri, 2004) . The details of our system are provided in \u00a75.", "cite_spans": [ { "start": 414, "end": 445, "text": "(Salomon, 1996\u037e Fe dorova, 2012", "ref_id": null }, { "start": 637, "end": 650, "text": "(Mohri, 2004)", "ref_id": "BIBREF30" } ], "ref_spans": [], "eq_spans": [], "section": "Introduction", "sec_num": "1" }, { "text": "The computational processing of Brahmic scripts is not a new topic, with the first applications dating back to the early formal syntactic work by Datta (1984) . With an increased focus on the South Asian languages within the NLP commu nity, facilitated by advances in machine learning and the increased availability of relevant corpora, multiple script processing solutions have emerged. Some of these toolkits, such as statistical ma chine translationbased BrahmiNet (Kunchukut tan et al., 2015) , are modelbased, while oth ers, such as URoman (Hermjakob et al., 2018) , IndicNLP (Kunchukuttan, 2020) , and Akshar mukha (Rajan, 2020), employ rules. The main fo cus of these libraries is script conversion and ro manization. In this capacity they were success fully employed in diverse downstream multilin gual NLP tasks such as neural machine transla tion (Zhang et al., 2020\u037e Amrhein and Sennrich, 2020) , morphological analysis (Hauer et al., 2019\u037e Murikinati et al., 2020 , named entity recogni tion (Huang et al., 2019) and partofspeech tag ging (Cardenas et al., 2019) .", "cite_spans": [ { "start": 146, "end": 158, "text": "Datta (1984)", "ref_id": "BIBREF9" }, { "start": 468, "end": 496, "text": "(Kunchukut tan et al., 2015)", "ref_id": null }, { "start": 545, "end": 569, "text": "(Hermjakob et al., 2018)", "ref_id": "BIBREF18" }, { "start": 581, "end": 601, "text": "(Kunchukuttan, 2020)", "ref_id": "BIBREF27" }, { "start": 857, "end": 889, "text": "(Zhang et al., 2020\u037e Amrhein and", "ref_id": null }, { "start": 890, "end": 905, "text": "Sennrich, 2020)", "ref_id": "BIBREF1" }, { "start": 931, "end": 975, "text": "(Hauer et al., 2019\u037e Murikinati et al., 2020", "ref_id": null }, { "start": 1004, "end": 1024, "text": "(Huang et al., 2019)", "ref_id": "BIBREF20" }, { "start": 1051, "end": 1074, "text": "(Cardenas et al., 2019)", "ref_id": "BIBREF5" } ], "ref_spans": [], "eq_spans": [], "section": "Related Work", "sec_num": "2" }, { "text": "Similar to the software mentioned above, our li brary does provide romanization, but unlike some of the packages, such as URoman, we guarantee reversibility from Latin back to the native script. Similar to others we do not focus on faithful in vertible transliteration of named entities which typically requires modelbased approaches (Se quiera et al., 2014) . Unlike the IndicNLP pack age, our software does not provide morphologi cal analysis, but instead offers significantly richer script normalization capabilities than other pack ages. These capabilities are functionally sepa rated into normalization to Normalization Form C (NFC) and visual normalization. Additionally, our library provides extensive scriptspecific well formedness grammars. Finally, in contrast to these other approaches, grammars in our library are maintained separately from the code for compila tion and application, allowing for maintenance of existing scripts and languages plus extension to new ones without having to modify any code. This is particularly important given that Unicode stan dards do change over time and there remain many languages left to cover.", "cite_spans": [ { "start": 334, "end": 358, "text": "(Se quiera et al., 2014)", "ref_id": null } ], "ref_spans": [], "eq_spans": [], "section": "Related Work", "sec_num": "2" }, { "text": "To the best of our knowledge this is the first publicly available general finitestate grammar ap proach for lowlevel processing of multiple Brah mic scripts since the early formal syntactic work by Datta (1984) and is the first such library de signed based on an observation by Sproat (2003) that the fundamental organizing principles of the Brahmic scripts can be algebraically formalized. In particular, all the core components of our li brary (inverse romanization, normalization and wellformedness) are compactly and efficiently represented as finite state transducers. Such for malization lends itself particularly well to runtime or offline integration with any finite state process ing pipeline, such as decoder components of in put methods (Ouyang et al., 2017\u037e Hellsten et al., 2017 ), text normalization for automatic speech recognition and texttospeech synthesis (Zhang et al., 2019) , among other natural language and speech applications. from the 3rd century BCE and fell out of use by the 5th century CE (Salomon, 1996\u037e Strauch, 2012\u037e Fedorova, 2012 . The main unit of lin ear graphemic representation in Brahmic scripts is known by its traditional Sanskritderived name aks \u0323ara. As Bright (1999) notes, it is often trans lated as \"syllable\" although it does not bear di rect correspondence to a syllable of speech, but rather to an orthographic syllable. The structure, or \"grammar\" of an aks \u0323ara is based on the follow ing common principles: an aks \u0323ara often consists of a consonant symbol , by default bearing an unmarked inherent vowel or attached diacritic (de pendent) vowel sign ( )\u037e but it may also be an independent vowel symbol , or a consonant sym bol with its inherent vowel \"muted\" by a special virama diacritic \u2205 ( \u2205 ). In any of these preceding scenarios, the base consonant can be replaced by a consonant cluster where all but the last conso nant lose their inherent vowel. When the individ ual component consonants of the cluster combine to form a composite form, precluding the use of an overt virama diacritic, this is known as a \"conso nant conjunct\" (e.g., 2013\u037e Bright, 1999\u037e Coulmas, 1999\u037e Share and Daniels, 2016 .", "cite_spans": [ { "start": 198, "end": 210, "text": "Datta (1984)", "ref_id": "BIBREF9" }, { "start": 278, "end": 291, "text": "Sproat (2003)", "ref_id": "BIBREF44" }, { "start": 748, "end": 791, "text": "(Ouyang et al., 2017\u037e Hellsten et al., 2017", "ref_id": "BIBREF17" }, { "start": 874, "end": 894, "text": "(Zhang et al., 2019)", "ref_id": "BIBREF51" }, { "start": 1018, "end": 1063, "text": "(Salomon, 1996\u037e Strauch, 2012\u037e Fedorova, 2012", "ref_id": null }, { "start": 1197, "end": 1210, "text": "Bright (1999)", "ref_id": "BIBREF4" }, { "start": 2094, "end": 2152, "text": "2013\u037e Bright, 1999\u037e Coulmas, 1999\u037e Share and Daniels, 2016", "ref_id": null } ], "ref_spans": [], "eq_spans": [], "section": "Related Work", "sec_num": "2" }, { "text": "\u2205 \u2205 vs [ ] 3 ) (Fe dorova,", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Brahmic Scripts: An Overview", "sec_num": "3" }, { "text": "The elements of the aks \u0323ara grammar described above can be grouped into several natural classes. The sizes of the core classes are shown in Ta ble 1 for each writing system and its correspond ing ISO 15924 identifier in uppercase format (ISO, 2004) . The major classes are the independent vow els (e.g., the Devanagari diphthong \u0914), the depen dent vowel diacritics (e.g., the Gujarati \u25cc\u0ac0), and the consonants (e.g., the Gurmukhi \u0a5c). Another im portant class consists of the coda consonant sym bols, like anusvara, chandrabindu, and visarga, which modify the aks \u0323ara as a whole (and follow and vowel signs in the memory representation). Fi nally, there is a class of special characters, such as the religious symbol Om \u0950, that behave like inde pendent aks \u0323ara. 4", "cite_spans": [ { "start": 238, "end": 249, "text": "(ISO, 2004)", "ref_id": "BIBREF22" } ], "ref_spans": [], "eq_spans": [], "section": "Brahmic Scripts: An Overview", "sec_num": "3" }, { "text": "Unicode Normalization Unicode defines sev eral normalization forms which are used for check ing whether the two Unicode strings are equiv alent to each other (Unicode Consortium, 2019) . In our library we support Normalization Form C (NFC) which is well suited for comparing visu ally identical strings. This normalization gener ally converts strings to the equivalent form that uses composite characters. Table 2 shows two ex amples of legacy sequences corresponding canon ically equivalent forms for Devanagari.", "cite_spans": [ { "start": 167, "end": 184, "text": "Consortium, 2019)", "ref_id": "BIBREF48" } ], "ref_spans": [ { "start": 406, "end": 413, "text": "Table 2", "ref_id": "TABREF3" } ], "eq_spans": [], "section": "Brahmic Scripts: An Overview", "sec_num": "3" }, { "text": "Visual Normalization As was mentioned above, an aks \u0323ara may be represented by multiple Unicode character sequences and the goal of NFC normal ization is to convert them to their unique canonical form. However, there are many Unicode character sequences that fall outside the scope of NFC algo rithm. We provide visual normalization that, in ad dition to providing the NFC functionality, also sup ports transforming such legacy sequences. Some of the rules are provided as \"Do Not Use\" tables by the Unicode Consortium (2019) that recommends transformations from legacy sequences to their cor responding canonical form, such as Devanagari { \u0905 (U+0905), \u0945 (U+0945) } \u2192 \u0972 (U+0972). We also included transformations for visually identical se quences (under many implementations) which are commonly found on the Web, such as Devanagari", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Brahmic Scripts: An Overview", "sec_num": "3" }, { "text": "{ \u0910 (U+0910), \u0947 (U+0947) } \u2192 \u0910 (U+0910). 5", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Brahmic Scripts: An Overview", "sec_num": "3" }, { "text": "Wellformedness Check A wellformedness ac ceptor verifies whether the given text is readable in a particular script or not. It would be hard for the native reader to visually parse the text if the script rules are not followed. For example, the reader does not expect two vowels signs on a single con sonant and such a thing may not even be possible to reasonably draw. Furthermore, unlike the Latin script, acronyms are not written using arbitrary let ter sequences, they are formed only as a sequence of aks \u0323ara. Our approach verifies whether the text is a sequence of wellformed aks \u0323ara using the gram mar described above.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Brahmic Scripts: An Overview", "sec_num": "3" }, { "text": "Reversible ISO Transliteration ISO 15919 rep resents a unified 8bit Latin transliteration scheme for major South Asian Brahmic scripts (ISO, 2001 ).", "cite_spans": [ { "start": 135, "end": 145, "text": "(ISO, 2001", "ref_id": "BIBREF21" } ], "ref_spans": [], "eq_spans": [], "section": "Brahmic Scripts: An Overview", "sec_num": "3" }, { "text": "Since it has not been updated with the characters that were introduced to the Unicode standard af ter 2001, we have added additional mappings, with some examples shown in Table 3 . These additions are crucial because they allow us to reverse the romanizations to get the original Brahmic strings back reliably. This property allows various data processing pipelines to use the romanized text as an internal representation and convert it back to the original native script at the output stage.", "cite_spans": [], "ref_spans": [ { "start": 171, "end": 178, "text": "Table 3", "ref_id": "TABREF5" } ], "eq_spans": [], "section": "Brahmic Scripts: An Overview", "sec_num": "3" }, { "text": "Languagespecific Logic Several South Asian languages often share the same script with some, often minor, languagespecific differences. Our library supports languagespecific customizations that can be combined with languageagnostic script logic. For example, the modern Bengali-Assamese script (Beng) is shared by both Bengali and Assamese languages, among others (Brandt and Sohoni, 2018) . For both of these languages our library provides customizations, 6 such as the transformations required for visual normal ization of Assamese that transform Bengali let ter ra into its Assamese equivalent when it par ticipates in a consonant conjunct (which gener ally occurs when following or preceding virama, e.g.,", "cite_spans": [ { "start": 363, "end": 388, "text": "(Brandt and Sohoni, 2018)", "ref_id": "BIBREF3" } ], "ref_spans": [], "eq_spans": [], "section": "Brahmic Scripts: An Overview", "sec_num": "3" }, { "text": "{ \u09b0 (U+09B0), \u25cc\u09cd (U+09CD) } \u2192 { \u09f0 (U+09F0), \u25cc\u09cd (U+09CD) }).", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Brahmic Scripts: An Overview", "sec_num": "3" }, { "text": "6 https://github.com/google-research/nisaba/ tree/main/nisaba/brahmic/data/lang Require: FSAs: consonant, vowel, vowel_sign, coda, standalone, virama, dead_consonant, accept. 1: function (consonant, vowel, vowel_sign, coda, standalone, virama, dead_consonant, accept) ", "cite_spans": [ { "start": 89, "end": 174, "text": "FSAs: consonant, vowel, vowel_sign, coda, standalone, virama, dead_consonant, accept.", "ref_id": null }, { "start": 187, "end": 267, "text": "(consonant, vowel, vowel_sign, coda, standalone, virama, dead_consonant, accept)", "ref_id": null } ], "ref_spans": [], "eq_spans": [], "section": "Brahmic Scripts: An Overview", "sec_num": "3" }, { "text": "The Brahmic script manipulation operations described above have a natural intepretation grounded in formal language theory. We treat the text corpus in a given script as a set of strings over some finite alphabet \u03a3 that defines a set of admissable script symbols. The set of zero or more strings is known as language which, in its simplest (regular) form, can be succintly described (or recognized) by a finite state automaton (FSA) or acceptor (Yu, 1997) . Two simple FSAs that represent the Gujarati word \u0aa6\u0ab8 are shown in Figure 1 , where the top automaton represents the word over an alphabet of Unicode code points for Gujarati, while the bottom one represents the same string over the corresponding byte symbols in UTF8 encoding (Unicode Consortium, 2019). Our library supports both representations. The aks \u0323ara grammar outlined in the previous section can be expressed via elementary formal op erations on the FSAs that describe grammar con stituents. Such settheoretic operations include union (\u222a), concatenation (+) and closure, where closure is defined as an arbitrary natural number of concatenations of a language over \u03a3 with it self, either accepting an empty input { } or not, denoted * (Kleene star) and + (Kleene plus), respectively (Kuich and Salomaa, 1986) . These operations represent nontrivial automata which are compiled offline resulting in compact and ef ficient representations. A simplified process for constructing the automaton to perform the well formed check from the previous section is shown in Figure 2 . In this simplified example, the paths through the automaton that define a legal conso nant cluster (line 2 of the algorithm) are repre sented by a subautomaton that recognizes the lan guage that consists of strings formed from the con sonant and virama symbols only, where each con sonant, apart from the last one, must be followed by the virama that removes an inherent vowel. The rest of the operations on the Brahmic scripts, namely the normalization and transliteration, in volve modifications of the Brahmic script inputs. Such operations are naturally expressed by finite state transducers (FSTs), which are a generaliza tion of the FSA concept used to encode string string relations (or transductions), by modifying the automata arcs to have pairs of labels from in put and output alphabets, instead of single labels. A trivial romanization in our representation of the two Sinhala words \u0d91\u0d9a (\u27e8eka\u27e9, \"one\") and \u0dd9\u0daf\u0d9a (\u27e8deka\u27e9, \"two\") is shown in Figure 3 . Note the \"vocalization\" of the final consonant by insertion of a schwa via an input transition. Also note that the path accepting the second word is longer. The word \u0dd9\u0daf\u0d9a consists of three aks \u0323ara and requires modification of the inherent vowel by the depen dent vowel in order to produce \u27e8de\u27e9.", "cite_spans": [ { "start": 445, "end": 455, "text": "(Yu, 1997)", "ref_id": "BIBREF50" }, { "start": 1248, "end": 1273, "text": "(Kuich and Salomaa, 1986)", "ref_id": "BIBREF25" } ], "ref_spans": [ { "start": 523, "end": 531, "text": "Figure 1", "ref_id": "FIGREF0" }, { "start": 1526, "end": 1534, "text": "Figure 2", "ref_id": "FIGREF1" }, { "start": 2486, "end": 2494, "text": "Figure 3", "ref_id": "FIGREF2" } ], "eq_spans": [], "section": "The FiniteState Approach", "sec_num": "4" }, { "text": "The basic operations on the FSAs outlined above also extend to the FST case and allow for similarly succinct final compiled representa tions (Mohri, 2000) , such as the simplified con struction of the ISO romanization transducer \u2110 for converting from Brahmic scripts to Latin alpha bet, shown in Figure 4 . An important extension of FSAs and FSTs are the weighted finite state au tomata (WFSAs) and transducers (WFSTs) (Mohri, 2004 (Mohri, , 2009 that equip each arc in the automaton or transducer with a weight, thus allowing optimiza tion and search algorithms to compute the costs of distinct paths, which can be used to determine their relative importance. We use weights in some of our grammars to indicate the relative priority of a par ticular aks \u0323ara modification. For example, in Fig ure 4 , the paths corresponding to consonants fol lowed by dependent vowels (line 6) have priority Require: FSTs: consonant, vowel, vowel_sign, coda, standalone, virama. 1: function \u2110 (consonant, vowel, vowel_sign, coda, standalone, virama) over the aks \u0323arainitial independent vowels (line 9). The two remaining operations on aks \u0323ara, namely NFC and visual normalization, are repre sented in our library using the contextdependent rewrite rules from the formal approach pop ularized by Chomsky and Halle (1968) . The normalization rules are represented as a sequence { \u2192 / __ }, where the source is rewritten as if its left and right contexts are and . For an earlier example from \u00a73, a single NFC normal ization rule rewrites the Devanagari string = \"\u0928\" (na, U+0928) + \"\u093c \" (nukta sign, U+093C) into its canonical composition = \"\u0929\" (nnna, U+0929). Kaplan and Kay (1994) proposed an algorithm for compiling such sequences into an FST. This approach was further improved and extended to WFSTs by Mohri and Sproat (1996) , whose algorithm we use to compile sequences of NFC and visual normalization rules into transducers denoted and . Finally, the transducers representing language specific customizations of a particular script op eration are compiled by composing the generic languageagnostic transducer, such as the Devana gari visual normalizer, with the transducer rep resenting transformations that capture language specific use of the script, e.g., Devanagari for Nepali.", "cite_spans": [ { "start": 141, "end": 154, "text": "(Mohri, 2000)", "ref_id": "BIBREF29" }, { "start": 419, "end": 431, "text": "(Mohri, 2004", "ref_id": "BIBREF30" }, { "start": 432, "end": 446, "text": "(Mohri, , 2009", "ref_id": "BIBREF31" }, { "start": 903, "end": 964, "text": "FSTs: consonant, vowel, vowel_sign, coda, standalone, virama.", "ref_id": null }, { "start": 979, "end": 1035, "text": "(consonant, vowel, vowel_sign, coda, standalone, virama)", "ref_id": null }, { "start": 1283, "end": 1307, "text": "Chomsky and Halle (1968)", "ref_id": "BIBREF7" }, { "start": 1646, "end": 1667, "text": "Kaplan and Kay (1994)", "ref_id": "BIBREF24" }, { "start": 1792, "end": 1815, "text": "Mohri and Sproat (1996)", "ref_id": "BIBREF32" } ], "ref_spans": [ { "start": 296, "end": 304, "text": "Figure 4", "ref_id": "FIGREF3" }, { "start": 790, "end": 800, "text": "Fig ure 4", "ref_id": "FIGREF3" } ], "eq_spans": [], "section": "The FiniteState Approach", "sec_num": "4" }, { "text": "The core of the Nisaba Brahmic script manipula tion library resides under the brahmic directory of the distribution. In this section we provide de tails for how to build and use the library and also explore its application to visual normalization of Wikipediabased text in 9 of these scripts.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "System Details and Demo", "sec_num": "5" }, { "text": "Prerequisites We use Bazel (Google, 2020) as a primary build environment. For compiling the Op. Symb.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "System Details and Demo", "sec_num": "5" }, { "text": "Prop. Script BENG DEVA GUJR GURU KNDA MLYM ORYA SINH TAML TELU 127 130 113 93 119 122 105 122 75 112 Unicode 475 546 476 418 487 522 452 513 326 485 248 235 195 171 210 201 178 192 126 181 \u2110 Byte 384 399 334 288 350 345 305 339 229 318 Unicode 9 17 1 8 21 8 9 17 11 4 158 248 75 78 349 261 160 352 228 163 Byte 31 55 1 28 70 27 31 55 37 14 1,812 1,841 255 1,047 2,884 2,322 1,813 2,611 3,098 1,543 103 51,710 98 119 1764 287 60 182 209 57 Unicode 2,423 121,157 2,234 2,322 6,136 3,021 1,732 2,129 1,280 2,249 369 165,168 356 425 5,611 965 232 624 703 225 Byte 18,896 266,441 18,684 20,733 30,422 18,598 16,146 15,363 11,830 18 automata and transducers we employ Pynini 7 , a Python library for constructing finitestate gram mars and for performing operations on WF STs (Gorman, 2016\u037e Gorman and Sproat, in press ).", "cite_spans": [ { "start": 688, "end": 754, "text": "18,896 266,441 18,684 20,733 30,422 18,598 16,146 15,363 11,830 18", "ref_id": null }, { "start": 897, "end": 939, "text": "(Gorman, 2016\u037e Gorman and Sproat, in press", "ref_id": null } ], "ref_spans": [ { "start": 6, "end": 687, "text": "Script BENG DEVA GUJR GURU KNDA MLYM ORYA SINH TAML TELU 127 130 113 93 119 122 105 122 75 112 Unicode 475 546 476 418 487 522 452 513 326 485 248 235 195 171 210 201 178 192 126 181 \u2110 Byte 384 399 334 288 350 345 305 339 229 318 Unicode 9 17 1 8 21 8 9 17 11 4 158 248 75 78 349 261 160 352 228 163 Byte 31 55 1 28 70 27 31 55 37 14 1,812 1,841 255 1,047 2,884 2,322 1,813 2,611 3,098 1,543 103 51,710 98 119 1764 287 60 182 209 57 Unicode 2,423 121,157 2,234 2,322 6,136 3,021 1,732 2,129 1,280 2,249 369 165,168 356 425 5,611 965 232 624 703 225 Byte", "ref_id": "TABREF1" } ], "eq_spans": [], "section": "System Details and Demo", "sec_num": "5" }, { "text": "In addition, the library depends on Thrax 8 , an older relative of Pynini, that provides a custom gram mar manipulation language for WFSTs (Tai et al., 2011\u037e Roark et al., 2012 . Although Thrax has been mostly superseded by Pynini, we still rely on some of its utilities for unit testing and its C++ run time components. At their core, both Pynini and Thrax depend on the OpenFst library 9 for the im plementation of most WFST algorithms (Allauzen et al., 2007\u037e Riley et al., 2009 . The overall depen dency diagram is shown on the lefthand side of Figure 5 (the minimal dependency on Thrax is in dicated by a dotted arrow). At build time, Bazel pulls in these dependencies remotely from their re spective repositories. Figure 6 presents the sequence of steps to compile the transduc ers, including downloading the repository (line 2), compiling the library and its artifacts (line 5) and running the unit tests (line 7). The artifacts are compiled by Bazel using Pynini and consist of the finite state archive (FAR) files that contain collec tions of WFSTs (Roark et al., 2012) . For each of the four Brahmic script operations we generate two FAR files: one for WFSTs over the byte al phabet, and another over the Unicode code point alphabet. 10 Each FAR file contains ten script specific transducers whose names correspond to the uppercase ISO 15924 script codes. Since the transliteration operation is bidirectional, the name of each scriptspecific transliteration transducer has the prefix FROM_ for the nativetoLatin direc tion, and TO_ for the inverse. The numbers of states ( ) and arcs ( ) of the resulting transliteration (\u2110), NFC ( ), visual normalization ( ) transduc ers and wellformedness acceptors ( ) for each script and alphabet type are shown in Table 4 .", "cite_spans": [ { "start": 139, "end": 176, "text": "(Tai et al., 2011\u037e Roark et al., 2012", "ref_id": null }, { "start": 438, "end": 480, "text": "(Allauzen et al., 2007\u037e Riley et al., 2009", "ref_id": null }, { "start": 1057, "end": 1077, "text": "(Roark et al., 2012)", "ref_id": "BIBREF38" } ], "ref_spans": [ { "start": 548, "end": 556, "text": "Figure 5", "ref_id": "FIGREF4" }, { "start": 719, "end": 727, "text": "Figure 6", "ref_id": null }, { "start": 1762, "end": 1769, "text": "Table 4", "ref_id": "TABREF9" } ], "eq_spans": [], "section": "System Details and Demo", "sec_num": "5" }, { "text": "Offline and Online Usage Once the transduc ers are compiled, they can be applied offline to the input files using the rewrite-tester tool pro vided by Thrax, as shown in lines 8-13 of the ex ample in Figure 6 , where the visual normalization transducer for Kannada that resides in the vi-sual_norm.far archive is applied to words in in put file words.txt. We provide lightweight runtime interfaces for iso_to_deva.ApplyOnText('\u27e8k\u02d1laba\u27e9')) # Check valid inputs. wellformed_mlym = brahmic.WellFormed('Mlym') self.assertTrue(wellformed_mlym.AcceptText('\u0d38\u0d4d\u0d35 \u0d30\u0d02')) # Visual normalizer. visual_norm_deva = brahmic.VisualNorm('Deva') self.assertEqual('\u0914', visual_norm_deva.ApplyOnText('\u0914')) both Python and C++, their dependencies shown in the center and the righthand side of Figure 5 , respectively. The Python interface is provided via several wrappers around the pynini.Fst abstrac tion, with a simple example shown in Figure 7 . In addition to performing simple operations on in dividual strings, more WFSTspecific operations, such as transducer composition, are provided by Pynini. The C++ interface is provided by the Grammar helper class, shown in Figure 8 , that includes the necessary methods for initializing the WFSTs and performing rewrites (for transducers) and ac ceptance tests (for acceptors). In addition, many more operations on WFSTs are available through the OpenFst library, if required.", "cite_spans": [], "ref_spans": [ { "start": 200, "end": 208, "text": "Figure 6", "ref_id": null }, { "start": 770, "end": 778, "text": "Figure 5", "ref_id": "FIGREF4" }, { "start": 916, "end": 924, "text": "Figure 7", "ref_id": "FIGREF5" }, { "start": 1149, "end": 1157, "text": "Figure 8", "ref_id": null } ], "eq_spans": [], "section": "Compiling the Transducers", "sec_num": null }, { "text": "To demonstrate the prevalence of text requiring normalization in #include // Generic wrapper around FST archive with Brahmic transducers. class Grammar { public: // Constructs given the FAR path, its name and the name of WFST. Grammar(const std::string& far_path, const std::string& far_name, const std::string& fst_name); // Initializes the transducer. bool Load(); // Rewrites into . bool Rewrite(const std::string& input, std::string *output) const; // Checks whether the grammar accepts . bool Accept(const std::string& input) const; }; Figure 8 : Runtime C++ interface. these scripts, we normalized publicly available cor pora and measured how frequently words in the samples were modified. The Dakshina dataset (Roark et al., 2020) includes (among other things) collections of monolingual Wikipedia sentences in 12 South Asian languages, 10 of which use Brah mic scripts. We applied visual normalization to the training partitions of the collections in these 10 lan guages, and Table 5 presents the percentage of both types and tokens that were changed by the normal ization. 11 Malayalam is the language with the high est percentage of both types and tokens changed by visual normalization, largely due to frequent con version to chillu letters from alternative encodings. For example, the relatively frequent word \u0d24\u0d46\u0d28\u0d4d\u0d31 (\"yours\") is normalized to the encoding with the chillu letter \u0d7b instead of \u0d28.", "cite_spans": [ { "start": 749, "end": 769, "text": "(Roark et al., 2020)", "ref_id": "BIBREF39" } ], "ref_spans": [ { "start": 573, "end": 581, "text": "Figure 8", "ref_id": null }, { "start": 1016, "end": 1023, "text": "Table 5", "ref_id": "TABREF12" } ], "eq_spans": [], "section": "Prevalence of Normalization", "sec_num": null }, { "text": "We presented finitestate automatabased utilities for processing the major Brahmic scripts. The fi nite state transducer formalism provides an effi cient and scalable framework for expressing Brah mic script operations and is suitable for many NLP applications, such as those reported in Kumar et al. (2020) and Kakwani et al. (2020) , which may ben efit from the reduction in \"noise\" present in unnor malized text. In the future, we will continue to im prove the support for existing scripts and extend our work to other Brahmic scripts.", "cite_spans": [ { "start": 287, "end": 306, "text": "Kumar et al. (2020)", "ref_id": "BIBREF26" }, { "start": 311, "end": 332, "text": "Kakwani et al. (2020)", "ref_id": null } ], "ref_spans": [], "eq_spans": [], "section": "Conclusion and Future Work", "sec_num": "6" }, { "text": "See \u00a73 for details on the scripts. 2 https://github.com/google-research/nisaba/", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "", "sec_num": null }, { "text": "Here, surrounding the consonants in square brackets will serve to indicate that the enclosed consonants form a conjunct together.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "", "sec_num": null }, { "text": "These classes are documented in https://github.com/ google-research/nisaba/blob/main/nisaba/brahmic/ mappings.md.5 Here the combining vowel sign U+0947 does not affect the compound glyph's visual appearance hence is removed.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "", "sec_num": null }, { "text": "http://pynini.opengrm.org/ 8 http://thrax.opengrm.org 9 http://www.openfst.org", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "", "sec_num": null }, { "text": "The Unicode code point FARs rather misleadingly have the suffix utf8 in their name for historical reasons.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "", "sec_num": null }, { "text": "Tokenization was simply based on whitespace, with no other processing such as punctuation separation, so the total number of distinct types is accordingly relatively high. The texts from that dataset were already NFC normalized.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "", "sec_num": null } ], "back_matter": [ { "text": "The authors would like to thank I\u015f\u0131n Demir\u015fahin for valuable discussion on this project.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Acknowledgments", "sec_num": null } ], "bib_entries": { "BIBREF0": { "ref_id": "b0", "title": "OpenFst: A general and efficient weighted finitestate transducer library", "authors": [ { "first": "Cyril", "middle": [], "last": "Allauzen", "suffix": "" }, { "first": "Michael", "middle": [], "last": "Riley", "suffix": "" }, { "first": "Johan", "middle": [], "last": "Schalkwyk", "suffix": "" } ], "year": 2007, "venue": "International Conference on Implemen tation and Application of Automata", "volume": "", "issue": "", "pages": "11--23", "other_ids": { "DOI": [ "10.1007/978-3-540-76336-9_3" ] }, "num": null, "urls": [], "raw_text": "Cyril Allauzen, Michael Riley, Johan Schalkwyk, Wo jciech Skut, and Mehryar Mohri. 2007. OpenFst: A general and efficient weighted finitestate transducer library. In International Conference on Implemen tation and Application of Automata, pages 11-23. Springer.", "links": null }, "BIBREF1": { "ref_id": "b1", "title": "On Roman ization for model transfer between scripts in neural machine translation", "authors": [ { "first": "Chantal", "middle": [], "last": "Amrhein", "suffix": "" }, { "first": "Rico", "middle": [], "last": "Sennrich", "suffix": "" } ], "year": 2020, "venue": "Findings of the Association for Computational Linguistics: EMNLP 2020", "volume": "", "issue": "", "pages": "2461--2469", "other_ids": { "DOI": [ "10.18653/v1/2020.findings-emnlp.223" ] }, "num": null, "urls": [], "raw_text": "Chantal Amrhein and Rico Sennrich. 2020. On Roman ization for model transfer between scripts in neural machine translation. In Findings of the Association for Computational Linguistics: EMNLP 2020, pages 2461-2469, Online. Association for Computational Linguistics.", "links": null }, "BIBREF2": { "ref_id": "b2", "title": "Indic language computing", "authors": [ { "first": "Pushpak", "middle": [], "last": "Bhattacharyya", "suffix": "" }, { "first": "Hema", "middle": [], "last": "Murthy", "suffix": "" }, { "first": "Surangika", "middle": [], "last": "Ranathunga", "suffix": "" }, { "first": "Ranjiva", "middle": [], "last": "Munasinghe", "suffix": "" } ], "year": 2019, "venue": "Communications of the ACM", "volume": "62", "issue": "11", "pages": "70--75", "other_ids": { "DOI": [ "10.1145/3343456" ] }, "num": null, "urls": [], "raw_text": "Pushpak Bhattacharyya, Hema Murthy, Surangika Ranathunga, and Ranjiva Munasinghe. 2019. Indic language computing. Communications of the ACM, 62(11):70-75.", "links": null }, "BIBREF3": { "ref_id": "b3", "title": "Script and identity -the politics of writing in South Asia: an in troduction", "authors": [ { "first": "Carmen", "middle": [], "last": "Brandt", "suffix": "" }, { "first": "Pushkar", "middle": [], "last": "Sohoni", "suffix": "" } ], "year": 2018, "venue": "South Asian History and Culture", "volume": "9", "issue": "1", "pages": "1--15", "other_ids": { "DOI": [ "10.1080/19472498.2017.1411048" ] }, "num": null, "urls": [], "raw_text": "Carmen Brandt and Pushkar Sohoni. 2018. Script and identity -the politics of writing in South Asia: an in troduction. South Asian History and Culture, 9(1):1- 15.", "links": null }, "BIBREF4": { "ref_id": "b4", "title": "A matter of typology: Alphasyl labaries and abugidas", "authors": [ { "first": "William", "middle": [], "last": "Bright", "suffix": "" } ], "year": 1999, "venue": "Written Language & Literacy", "volume": "2", "issue": "", "pages": "45--55", "other_ids": { "DOI": [ "10.1075/wll.2.1.03bri" ] }, "num": null, "urls": [], "raw_text": "William Bright. 1999. A matter of typology: Alphasyl labaries and abugidas. Written Language & Literacy, 2(1):45-55.", "links": null }, "BIBREF5": { "ref_id": "b5", "title": "A grounded unsupervised universal partof speech tagger for lowresource languages", "authors": [ { "first": "Ronald", "middle": [], "last": "Cardenas", "suffix": "" }, { "first": "Ying", "middle": [], "last": "Lin", "suffix": "" }, { "first": "Ji", "middle": [], "last": "Heng", "suffix": "" }, { "first": "Jonathan", "middle": [], "last": "", "suffix": "" } ], "year": 2019, "venue": "Pro ceedings of the 2019 Conference of the North Amer ican Chapter of the Association for Computational Linguistics: Human Language Technologies", "volume": "1", "issue": "", "pages": "2428--2439", "other_ids": { "DOI": [ "10.18653/v1/N19-1252" ] }, "num": null, "urls": [], "raw_text": "Ronald Cardenas, Ying Lin, Heng Ji, and Jonathan May. 2019. A grounded unsupervised universal partof speech tagger for lowresource languages. In Pro ceedings of the 2019 Conference of the North Amer ican Chapter of the Association for Computational Linguistics: Human Language Technologies, Vol ume 1 (Long and Short Papers), pages 2428-2439, Minneapolis, Minnesota. Association for Computa tional Linguistics.", "links": null }, "BIBREF6": { "ref_id": "b6", "title": "The Indo Aryan Languages. Routledge Language Family Se ries. Routledge", "authors": [ { "first": "George", "middle": [], "last": "Cardona", "suffix": "" }, { "first": "Danesh", "middle": [], "last": "Jain", "suffix": "" } ], "year": 2007, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "George Cardona and Danesh Jain. 2007. The Indo Aryan Languages. Routledge Language Family Se ries. Routledge, New York.", "links": null }, "BIBREF7": { "ref_id": "b7", "title": "The Sound Pattern of English", "authors": [ { "first": "Noam", "middle": [], "last": "Chomsky", "suffix": "" }, { "first": "Morris", "middle": [], "last": "Halle", "suffix": "" } ], "year": 1968, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "Noam Chomsky and Morris Halle. 1968. The Sound Pattern of English. Harper & Row, New York.", "links": null }, "BIBREF8": { "ref_id": "b8", "title": "The Blackwell Encyclopedia of Writing Systems", "authors": [ { "first": "Florian", "middle": [], "last": "Coulmas", "suffix": "" } ], "year": 1999, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": { "DOI": [ "10.1002/9781118932667" ] }, "num": null, "urls": [], "raw_text": "Florian Coulmas. 1999. The Blackwell Encyclopedia of Writing Systems. John Wiley & Sons, Oxford.", "links": null }, "BIBREF9": { "ref_id": "b9", "title": "A generalized formal approach for description and analysis of major Indian scripts", "authors": [ { "first": "A", "middle": [ "K" ], "last": "Datta", "suffix": "" } ], "year": 1984, "venue": "IETE Journal of Research", "volume": "30", "issue": "6", "pages": "155--161", "other_ids": { "DOI": [ "10.1080/03772063.1984.11453262" ] }, "num": null, "urls": [], "raw_text": "A. K. Datta. 1984. A generalized formal approach for description and analysis of major Indian scripts. IETE Journal of Research, 30(6):155-161.", "links": null }, "BIBREF10": { "ref_id": "b10", "title": "The development of structural characteristics of Brahmi script in deriva tive writing systems", "authors": [ { "first": "L", "middle": [], "last": "Liudmila", "suffix": "" }, { "first": "", "middle": [], "last": "Fedorova", "suffix": "" } ], "year": 2012, "venue": "Written Language & Literacy", "volume": "15", "issue": "1", "pages": "1--25", "other_ids": { "DOI": [ "10.1075/wll.15.1.01fed" ] }, "num": null, "urls": [], "raw_text": "Liudmila L Fedorova. 2012. The development of structural characteristics of Brahmi script in deriva tive writing systems. Written Language & Literacy, 15(1):1-25.", "links": null }, "BIBREF11": { "ref_id": "b11", "title": "The development of graphic representation in abugida writing: The ak shara's grammar", "authors": [ { "first": "L", "middle": [], "last": "Liudmila", "suffix": "" }, { "first": "", "middle": [], "last": "Fedorova", "suffix": "" } ], "year": 2013, "venue": "Lingua Posnaniensis", "volume": "55", "issue": "2", "pages": "49--66", "other_ids": { "DOI": [ "10.2478/linpo-2013-0013" ] }, "num": null, "urls": [], "raw_text": "Liudmila L. Fedorova. 2013. The development of graphic representation in abugida writing: The ak shara's grammar. Lingua Posnaniensis, 55(2):49- 66.", "links": null }, "BIBREF13": { "ref_id": "b13", "title": "Pynini: A Python library for weighted finitestate grammar compilation", "authors": [ { "first": "Kyle", "middle": [], "last": "Gorman", "suffix": "" } ], "year": 2016, "venue": "Pro ceedings of the SIGFSM Workshop on Statistical NLP and Weighted Automata", "volume": "", "issue": "", "pages": "75--80", "other_ids": { "DOI": [ "10.18653/v1/W16-2409" ] }, "num": null, "urls": [], "raw_text": "Kyle Gorman. 2016. Pynini: A Python library for weighted finitestate grammar compilation. In Pro ceedings of the SIGFSM Workshop on Statistical NLP and Weighted Automata, pages 75-80, Berlin, Germany. Association for Computational Linguis tics.", "links": null }, "BIBREF14": { "ref_id": "b14", "title": "press. FiniteState Text Processing. Human Language Technologies", "authors": [ { "first": "Kyle", "middle": [], "last": "Gorman", "suffix": "" }, { "first": "Richard", "middle": [], "last": "Sproat", "suffix": "" } ], "year": null, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "Kyle Gorman and Richard Sproat. in press. FiniteState Text Processing. Human Language Technologies.", "links": null }, "BIBREF16": { "ref_id": "b16", "title": "Cognate projection for lowresource inflection gen eration", "authors": [ { "first": "Bradley", "middle": [], "last": "Hauer", "suffix": "" }, { "first": "Ahmad", "middle": [], "last": "Amir", "suffix": "" }, { "first": "Yixing", "middle": [], "last": "Habibi", "suffix": "" }, { "first": "", "middle": [], "last": "Luan", "suffix": "" } ], "year": 2019, "venue": "Proceedings of the 16th Workshop on Computational Research in Phonetics, Phonology, and Morphology", "volume": "", "issue": "", "pages": "6--11", "other_ids": { "DOI": [ "10.18653/v1/W19-4202" ] }, "num": null, "urls": [], "raw_text": "Bradley Hauer, Amir Ahmad Habibi, Yixing Luan, Rashed Rubby Riyadh, and Grzegorz Kondrak. 2019. Cognate projection for lowresource inflection gen eration. In Proceedings of the 16th Workshop on Computational Research in Phonetics, Phonology, and Morphology, pages 6-11, Florence, Italy. Asso ciation for Computational Linguistics.", "links": null }, "BIBREF17": { "ref_id": "b17", "title": "Transliterated mo bile keyboard input via weighted finitestate trans ducers", "authors": [ { "first": "Lars", "middle": [], "last": "Hellsten", "suffix": "" }, { "first": "Brian", "middle": [], "last": "Roark", "suffix": "" }, { "first": "Prasoon", "middle": [], "last": "Goyal", "suffix": "" }, { "first": "Cyril", "middle": [], "last": "Al Lauzen", "suffix": "" }, { "first": "Fran\u00e7oise", "middle": [], "last": "Beaufays", "suffix": "" }, { "first": "Tom", "middle": [], "last": "Ouyang", "suffix": "" }, { "first": "Michael", "middle": [], "last": "Riley", "suffix": "" }, { "first": "David", "middle": [], "last": "Rybach", "suffix": "" } ], "year": 2017, "venue": "Proceedings of the 13th International Conference on Finite State Methods and Natural Language Processing", "volume": "", "issue": "", "pages": "10--19", "other_ids": { "DOI": [ "10.18653/v1/W17-4002" ] }, "num": null, "urls": [], "raw_text": "Lars Hellsten, Brian Roark, Prasoon Goyal, Cyril Al lauzen, Fran\u00e7oise Beaufays, Tom Ouyang, Michael Riley, and David Rybach. 2017. Transliterated mo bile keyboard input via weighted finitestate trans ducers. In Proceedings of the 13th International Conference on Finite State Methods and Natural Language Processing (FSMNLP 2017), pages 10- 19, Ume\u00e5, Sweden. Association for Computational Linguistics.", "links": null }, "BIBREF18": { "ref_id": "b18", "title": "Outofthebox universal Romanization tool uroman", "authors": [ { "first": "Ulf", "middle": [], "last": "Hermjakob", "suffix": "" }, { "first": "Jonathan", "middle": [], "last": "May", "suffix": "" }, { "first": "Kevin", "middle": [], "last": "Knight", "suffix": "" } ], "year": 2018, "venue": "Proceedings of ACL 2018, System Demonstrations", "volume": "", "issue": "", "pages": "13--18", "other_ids": { "DOI": [ "10.18653/v1/P18-4003" ] }, "num": null, "urls": [], "raw_text": "Ulf Hermjakob, Jonathan May, and Kevin Knight. 2018. Outofthebox universal Romanization tool uroman. In Proceedings of ACL 2018, System Demonstrations, pages 13-18, Melbourne, Australia. Association for Computational Linguistics.", "links": null }, "BIBREF19": { "ref_id": "b19", "title": "Design and evaluation of soft keyboards for Brahmic scripts", "authors": [ { "first": "Lauren", "middle": [], "last": "Hinkle", "suffix": "" }, { "first": "Albert", "middle": [], "last": "Brouillette", "suffix": "" }, { "first": "Sujay", "middle": [], "last": "Jayakar", "suffix": "" }, { "first": "Leigh", "middle": [], "last": "Gathings", "suffix": "" }, { "first": "Miguel", "middle": [], "last": "Lezcano", "suffix": "" }, { "first": "Jugal", "middle": [], "last": "Kalita", "suffix": "" } ], "year": 2013, "venue": "ACM Transactions on Asian Language Infor mation Processing (TALIP)", "volume": "12", "issue": "2", "pages": "1--37", "other_ids": { "DOI": [ "10.1145/2461316.2461318" ] }, "num": null, "urls": [], "raw_text": "Lauren Hinkle, Albert Brouillette, Sujay Jayakar, Leigh Gathings, Miguel Lezcano, and Jugal Kalita. 2013. Design and evaluation of soft keyboards for Brahmic scripts. ACM Transactions on Asian Language Infor mation Processing (TALIP), 12(2):1-37.", "links": null }, "BIBREF20": { "ref_id": "b20", "title": "What matters for neural crosslingual named entity recognition: An empirical analysis", "authors": [ { "first": "Xiaolei", "middle": [], "last": "Huang", "suffix": "" }, { "first": "Jonathan", "middle": [], "last": "May", "suffix": "" }, { "first": "Nanyun", "middle": [], "last": "Peng", "suffix": "" } ], "year": 2019, "venue": "Proceedings of the 2019 Conference on Empirical Methods in Nat ural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLPIJCNLP)", "volume": "", "issue": "", "pages": "6395--6401", "other_ids": { "DOI": [ "10.18653/v1/D19-1672" ] }, "num": null, "urls": [], "raw_text": "Xiaolei Huang, Jonathan May, and Nanyun Peng. 2019. What matters for neural crosslingual named entity recognition: An empirical analysis. In Proceedings of the 2019 Conference on Empirical Methods in Nat ural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLPIJCNLP), pages 6395-6401, Hong Kong, China. Association for Computational Linguistics.", "links": null }, "BIBREF21": { "ref_id": "b21", "title": "ISO 15919: Transliteration of Devana gari and related Indic scripts into Latin characters", "authors": [ { "first": "", "middle": [], "last": "Iso", "suffix": "" } ], "year": 2001, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "ISO. 2001. ISO 15919: Transliteration of Devana gari and related Indic scripts into Latin characters. https://www.iso.org/standard/28333.html. Interna tional Organization for Standardization.", "links": null }, "BIBREF22": { "ref_id": "b22", "title": "ISO 15924: Codes for the representation of names of scripts", "authors": [ { "first": "", "middle": [], "last": "Iso", "suffix": "" } ], "year": 2004, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "ISO. 2004. ISO 15924: Codes for the representation of names of scripts. https://www.iso.org/obp/ui/#iso: std:iso:15924:ed-1:v1:en. International Organiza tion for Standardization.", "links": null }, "BIBREF23": { "ref_id": "b23", "title": "2020. iNLPSuite: Monolingual corpora, evaluation benchmarks and pretrained multilingual language models for Indian languages", "authors": [ { "first": "Divyanshu", "middle": [], "last": "Kakwani", "suffix": "" }, { "first": "Anoop", "middle": [], "last": "Kunchukuttan", "suffix": "" }, { "first": "Satish", "middle": [], "last": "Golla", "suffix": "" }, { "first": "N", "middle": [ "C" ], "last": "Gokul", "suffix": "" }, { "first": "Avik", "middle": [], "last": "Bhattacharyya", "suffix": "" }, { "first": "M", "middle": [], "last": "Mitesh", "suffix": "" }, { "first": "Pratyush", "middle": [], "last": "Khapra", "suffix": "" }, { "first": "", "middle": [], "last": "Kumar", "suffix": "" } ], "year": null, "venue": "Proc. of the 2020 Conference on Em pirical Methods in Natural Language Processing: Findings, EMNLP 2020", "volume": "", "issue": "", "pages": "4948--4961", "other_ids": {}, "num": null, "urls": [], "raw_text": "Divyanshu Kakwani, Anoop Kunchukuttan, Satish Golla, Gokul N. C., Avik Bhattacharyya, Mitesh M. Khapra, and Pratyush Kumar. 2020. iNLPSuite: Monolingual corpora, evaluation benchmarks and pretrained multilingual language models for Indian languages. In Proc. of the 2020 Conference on Em pirical Methods in Natural Language Processing: Findings, EMNLP 2020, pages 4948-4961, Online Event. Association for Computational Linguistics.", "links": null }, "BIBREF24": { "ref_id": "b24", "title": "Regular mod els of phonological rule systems", "authors": [ { "first": "M", "middle": [], "last": "Ronald", "suffix": "" }, { "first": "Martin", "middle": [], "last": "Kaplan", "suffix": "" }, { "first": "", "middle": [], "last": "Kay", "suffix": "" } ], "year": 1994, "venue": "Computational Linguistics", "volume": "20", "issue": "3", "pages": "331--378", "other_ids": {}, "num": null, "urls": [], "raw_text": "Ronald M. Kaplan and Martin Kay. 1994. Regular mod els of phonological rule systems. Computational Linguistics, 20(3):331-378.", "links": null }, "BIBREF25": { "ref_id": "b25", "title": "Semirings, Au tomata, Languages, volume 5 of Monographs in The oretical Computer Science", "authors": [ { "first": "Werner", "middle": [], "last": "Kuich", "suffix": "" }, { "first": "Arto", "middle": [], "last": "Salomaa", "suffix": "" } ], "year": 1986, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": { "DOI": [ "10.1007/978-3-642-69959-7" ] }, "num": null, "urls": [], "raw_text": "Werner Kuich and Arto Salomaa. 1986. Semirings, Au tomata, Languages, volume 5 of Monographs in The oretical Computer Science. Springer, Berlin.", "links": null }, "BIBREF26": { "ref_id": "b26", "title": "A passage to In dia\": Pretrained word embeddings for Indian lan guages", "authors": [ { "first": "Saurav", "middle": [], "last": "Kumar", "suffix": "" }, { "first": "Saunack", "middle": [], "last": "Kumar", "suffix": "" }, { "first": "Diptesh", "middle": [], "last": "Kanojia", "suffix": "" }, { "first": "Pushpak", "middle": [], "last": "Bhattacharyya", "suffix": "" } ], "year": 2020, "venue": "Proc. of the 1st Joint Workshop on Spoken Language Technologies for Underresourced lan guages (SLTU) and Collaboration and Computing for UnderResourced Languages (CCURL)", "volume": "", "issue": "", "pages": "352--357", "other_ids": {}, "num": null, "urls": [], "raw_text": "Saurav Kumar, Saunack Kumar, Diptesh Kanojia, and Pushpak Bhattacharyya. 2020. \"A passage to In dia\": Pretrained word embeddings for Indian lan guages. In Proc. of the 1st Joint Workshop on Spoken Language Technologies for Underresourced lan guages (SLTU) and Collaboration and Computing for UnderResourced Languages (CCURL), pages 352-357, Marseille, France. European Language Re sources association.", "links": null }, "BIBREF27": { "ref_id": "b27", "title": "The IndicNLP Li brary", "authors": [ { "first": "Anoop", "middle": [], "last": "Kunchukuttan", "suffix": "" } ], "year": 2020, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "Anoop Kunchukuttan. 2020. The IndicNLP Li brary. https://github.com/anoopkunchukuttan/ indic_nlp_library.", "links": null }, "BIBREF28": { "ref_id": "b28", "title": "Brahminet: A transliteration and script conversion system for languages of the In dian subcontinent", "authors": [ { "first": "Anoop", "middle": [], "last": "Kunchukuttan", "suffix": "" }, { "first": "Ratish", "middle": [], "last": "Puduppully", "suffix": "" }, { "first": "Pushpak", "middle": [], "last": "Bhattacharyya", "suffix": "" } ], "year": 2015, "venue": "Proceedings of the 2015 Con ference of the North American Chapter of the Asso ciation for Computational Linguistics: Demonstra tions", "volume": "", "issue": "", "pages": "81--85", "other_ids": { "DOI": [ "10.3115/v1/N15-3017" ] }, "num": null, "urls": [], "raw_text": "Anoop Kunchukuttan, Ratish Puduppully, and Pushpak Bhattacharyya. 2015. Brahminet: A transliteration and script conversion system for languages of the In dian subcontinent. In Proceedings of the 2015 Con ference of the North American Chapter of the Asso ciation for Computational Linguistics: Demonstra tions, pages 81-85, Denver, Colorado. Association for Computational Linguistics.", "links": null }, "BIBREF29": { "ref_id": "b29", "title": "Minimization algorithms for se quential transducers", "authors": [ { "first": "Mehryar", "middle": [], "last": "Mohri", "suffix": "" } ], "year": 2000, "venue": "Theoretical Computer Science", "volume": "234", "issue": "12", "pages": "177--201", "other_ids": { "DOI": [ "10.1016/S0304-3975(98)00115-7" ] }, "num": null, "urls": [], "raw_text": "Mehryar Mohri. 2000. Minimization algorithms for se quential transducers. Theoretical Computer Science, 234(12):177-201.", "links": null }, "BIBREF30": { "ref_id": "b30", "title": "Weighted finitestate transducer algorithms. An overview", "authors": [ { "first": "Mehryar", "middle": [], "last": "Mohri", "suffix": "" } ], "year": 2004, "venue": "For mal Languages and Applications", "volume": "", "issue": "", "pages": "551--563", "other_ids": { "DOI": [ "10.1007/978-3-540-39886-8_29" ] }, "num": null, "urls": [], "raw_text": "Mehryar Mohri. 2004. Weighted finitestate transducer algorithms. An overview. In Carlos Mart\u00ednVide, Victor Mitrana, and Gheorghe P\u0103un, editors, For mal Languages and Applications, pages 551-563. Springer, Berlin\u037e Heidelberg.", "links": null }, "BIBREF31": { "ref_id": "b31", "title": "Weighted automata algorithms", "authors": [ { "first": "Mehryar", "middle": [], "last": "Mohri", "suffix": "" } ], "year": 2009, "venue": "Handbook of Weighted Automata, Mono graphs in Theoretical Computer Science", "volume": "", "issue": "", "pages": "213--254", "other_ids": { "DOI": [ "10.1007/978-3-642-01492-5_6" ] }, "num": null, "urls": [], "raw_text": "Mehryar Mohri. 2009. Weighted automata algorithms. In Manfred Droste, Werner Kuich, and Heiko Vogler, editors, Handbook of Weighted Automata, Mono graphs in Theoretical Computer Science, pages 213- 254. Springer.", "links": null }, "BIBREF32": { "ref_id": "b32", "title": "An efficient compiler for weighted rewrite rules", "authors": [ { "first": "Mehryar", "middle": [], "last": "Mohri", "suffix": "" }, { "first": "Richard", "middle": [], "last": "Sproat", "suffix": "" } ], "year": 1996, "venue": "34th An nual Meeting of the Association for Computational Linguistics", "volume": "", "issue": "", "pages": "231--238", "other_ids": { "DOI": [ "10.3115/981863.981894" ] }, "num": null, "urls": [], "raw_text": "Mehryar Mohri and Richard Sproat. 1996. An efficient compiler for weighted rewrite rules. In 34th An nual Meeting of the Association for Computational Linguistics, pages 231-238, Santa Cruz, California, USA. Association for Computational Linguistics.", "links": null }, "BIBREF33": { "ref_id": "b33", "title": "Transliteration for crosslingual morphological inflection", "authors": [ { "first": "Nikitha", "middle": [], "last": "Murikinati", "suffix": "" }, { "first": "Antonios", "middle": [], "last": "Anastasopoulos", "suffix": "" }, { "first": "Gra", "middle": [], "last": "Ham Neubig", "suffix": "" } ], "year": 2020, "venue": "Proceedings of the 17th SIGMORPHON Workshop on Computational Research in Phonetics, Phonology, and Morphology", "volume": "", "issue": "", "pages": "189--197", "other_ids": { "DOI": [ "10.18653/v1/2020.sigmorphon-1.22" ] }, "num": null, "urls": [], "raw_text": "Nikitha Murikinati, Antonios Anastasopoulos, and Gra ham Neubig. 2020. Transliteration for crosslingual morphological inflection. In Proceedings of the 17th SIGMORPHON Workshop on Computational Research in Phonetics, Phonology, and Morphology, pages 189-197, Online. Association for Computa tional Linguistics.", "links": null }, "BIBREF34": { "ref_id": "b34", "title": "Mobile keyboard input decod ing with finitestate transducers", "authors": [ { "first": "Tom", "middle": [], "last": "Ouyang", "suffix": "" }, { "first": "David", "middle": [], "last": "Rybach", "suffix": "" }, { "first": "Fran\u00e7oise", "middle": [], "last": "Beaufays", "suffix": "" }, { "first": "Michael", "middle": [], "last": "Riley", "suffix": "" } ], "year": 2017, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "Tom Ouyang, David Rybach, Fran\u00e7oise Beaufays, and Michael Riley. 2017. Mobile keyboard input decod ing with finitestate transducers.", "links": null }, "BIBREF35": { "ref_id": "b35", "title": "Handwriting recognition in Indian re gional scripts: A survey of offline techniques", "authors": [ { "first": "Umapada", "middle": [], "last": "Pal", "suffix": "" }, { "first": "Nabin", "middle": [], "last": "Ramachandran Jayadevan", "suffix": "" }, { "first": "", "middle": [], "last": "Sharma", "suffix": "" } ], "year": 2012, "venue": "ACM Transactions on Asian Language Information Pro cessing (TALIP)", "volume": "11", "issue": "1", "pages": "1--35", "other_ids": { "DOI": [ "http://doi.acm.org/10.1145/2090176.2090177" ] }, "num": null, "urls": [], "raw_text": "Umapada Pal, Ramachandran Jayadevan, and Nabin Sharma. 2012. Handwriting recognition in Indian re gional scripts: A survey of offline techniques. ACM Transactions on Asian Language Information Pro cessing (TALIP), 11(1):1-35.", "links": null }, "BIBREF37": { "ref_id": "b37", "title": "OpenFst: An opensource, weighted finite state transducer library and its applications to speech and language", "authors": [ { "first": "Michael", "middle": [], "last": "Riley", "suffix": "" }, { "first": "Cyril", "middle": [], "last": "Allauzen", "suffix": "" }, { "first": "Martin", "middle": [], "last": "Jansche", "suffix": "" } ], "year": 2009, "venue": "Proceedings of Human Language Technologies: The 2009 Annual Conference of the North American Chapter of the Association for Com putational Linguistics, Companion Volume: Tutorial Abstracts", "volume": "", "issue": "", "pages": "9--10", "other_ids": {}, "num": null, "urls": [], "raw_text": "Michael Riley, Cyril Allauzen, and Martin Jansche. 2009. OpenFst: An opensource, weighted finite state transducer library and its applications to speech and language. In Proceedings of Human Language Technologies: The 2009 Annual Conference of the North American Chapter of the Association for Com putational Linguistics, Companion Volume: Tutorial Abstracts, pages 9-10, Boulder, Colorado. Associa tion for Computational Linguistics.", "links": null }, "BIBREF38": { "ref_id": "b38", "title": "The OpenGrm opensource finitestate grammar soft ware libraries", "authors": [ { "first": "Brian", "middle": [], "last": "Roark", "suffix": "" }, { "first": "Richard", "middle": [], "last": "Sproat", "suffix": "" }, { "first": "Cyril", "middle": [], "last": "Allauzen", "suffix": "" }, { "first": "Michael", "middle": [], "last": "Riley", "suffix": "" }, { "first": "Jeffrey", "middle": [], "last": "Sorensen", "suffix": "" }, { "first": "Terry", "middle": [], "last": "Tai", "suffix": "" } ], "year": 2012, "venue": "Proceedings of the ACL 2012 Sys tem Demonstrations", "volume": "", "issue": "", "pages": "61--66", "other_ids": {}, "num": null, "urls": [], "raw_text": "Brian Roark, Richard Sproat, Cyril Allauzen, Michael Riley, Jeffrey Sorensen, and Terry Tai. 2012. The OpenGrm opensource finitestate grammar soft ware libraries. In Proceedings of the ACL 2012 Sys tem Demonstrations, pages 61-66, Jeju Island, Ko rea. Association for Computational Linguistics.", "links": null }, "BIBREF39": { "ref_id": "b39", "title": "Processing South Asian languages written in the Latin script: the Dakshina dataset", "authors": [ { "first": "Brian", "middle": [], "last": "Roark", "suffix": "" }, { "first": "Lawrence", "middle": [], "last": "Wolfsonkin", "suffix": "" }, { "first": "Christo", "middle": [], "last": "Kirov", "suffix": "" }, { "first": "Sabrina", "middle": [ "J" ], "last": "Mielke", "suffix": "" }, { "first": "Cibu", "middle": [], "last": "Johny", "suffix": "" }, { "first": "Isin", "middle": [], "last": "Demirsahin", "suffix": "" }, { "first": "Keith", "middle": [], "last": "Hall", "suffix": "" } ], "year": 2020, "venue": "Proc. of 12th Language Resources and Evaluation Conference (LREC)", "volume": "", "issue": "", "pages": "2413--2423", "other_ids": {}, "num": null, "urls": [], "raw_text": "Brian Roark, Lawrence WolfSonkin, Christo Kirov, Sabrina J. Mielke, Cibu Johny, Isin Demirsahin, and Keith Hall. 2020. Processing South Asian languages written in the Latin script: the Dakshina dataset. In Proc. of 12th Language Resources and Evaluation Conference (LREC), pages 2413-2423, Marseille, France.", "links": null }, "BIBREF40": { "ref_id": "b40", "title": "Brahmi and Kharoshthi", "authors": [ { "first": "G", "middle": [], "last": "Richard", "suffix": "" }, { "first": "", "middle": [], "last": "Salomon", "suffix": "" } ], "year": 1996, "venue": "The World's Writing Systems", "volume": "", "issue": "", "pages": "373--383", "other_ids": {}, "num": null, "urls": [], "raw_text": "Richard G. Salomon. 1996. Brahmi and Kharoshthi. In Peter T. Daniels and William Bright, editors, The World's Writing Systems, pages 373-383. Oxford University Press, New York, NY.", "links": null }, "BIBREF41": { "ref_id": "b41", "title": "Wordlevel language identifica tion and back transliteration of romanized text", "authors": [ { "first": "Shashank", "middle": [ "S" ], "last": "Royal Denzil Sequiera", "suffix": "" }, { "first": "B", "middle": [ "R" ], "last": "Rao", "suffix": "" }, { "first": "", "middle": [], "last": "Shambavi", "suffix": "" } ], "year": 2014, "venue": "Proceedings of the Forum for Information Retrieval Evaluation", "volume": "", "issue": "", "pages": "70--73", "other_ids": { "DOI": [ "10.1145/2824864.2824871" ] }, "num": null, "urls": [], "raw_text": "Royal Denzil Sequiera, Shashank S. Rao, and B. R. Shambavi. 2014. Wordlevel language identifica tion and back transliteration of romanized text. In Proceedings of the Forum for Information Retrieval Evaluation, pages 70-73, Bangalore, India.", "links": null }, "BIBREF42": { "ref_id": "b42", "title": "Aksha ras, alphasyllabaries, abugidas, alphabets and ortho graphic depth: Reflections on Rimzhim", "authors": [ { "first": "L", "middle": [], "last": "David", "suffix": "" }, { "first": "Peter", "middle": [ "T" ], "last": "Share", "suffix": "" }, { "first": "", "middle": [], "last": "Daniels", "suffix": "" } ], "year": 2014, "venue": "", "volume": "8", "issue": "", "pages": "17--31", "other_ids": { "DOI": [ "10.1080/17586801.2015.1016395" ] }, "num": null, "urls": [], "raw_text": "David L. Share and Peter T. Daniels. 2016. Aksha ras, alphasyllabaries, abugidas, alphabets and ortho graphic depth: Reflections on Rimzhim, Katz and Fowler (2014). Writing systems research, 8(1):17- 31.", "links": null }, "BIBREF43": { "ref_id": "b43", "title": "A journey from Indian scripts processing to Indian language processing", "authors": [ { "first": "R", "middle": [], "last": "", "suffix": "" }, { "first": "Mahesh", "middle": [ "K" ], "last": "Sinha", "suffix": "" } ], "year": 2009, "venue": "IEEE Annals of the History of Computing", "volume": "31", "issue": "1", "pages": "8--31", "other_ids": { "DOI": [ "10.1109/MAHC.2009.1" ] }, "num": null, "urls": [], "raw_text": "R. Mahesh K. Sinha. 2009. A journey from Indian scripts processing to Indian language processing. IEEE Annals of the History of Computing, 31(1):8- 31.", "links": null }, "BIBREF44": { "ref_id": "b44", "title": "A formal computational analysis of Indic scripts", "authors": [ { "first": "Richard", "middle": [], "last": "Sproat", "suffix": "" } ], "year": 2003, "venue": "International Symposium on Indic Scripts: Past and Future", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "Richard Sproat. 2003. A formal computational analysis of Indic scripts. In In International Symposium on Indic Scripts: Past and Future, Tokyo, Japan.", "links": null }, "BIBREF45": { "ref_id": "b45", "title": "Routledge Language Family Series", "authors": [ { "first": "B", "middle": [], "last": "Sanford", "suffix": "" }, { "first": "", "middle": [], "last": "Steever", "suffix": "" } ], "year": 2019, "venue": "The Dravidian Languages", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "Sanford B. Steever. 2019. The Dravidian Languages, 2nd edition. Routledge Language Family Series. Routledge, New York.", "links": null }, "BIBREF46": { "ref_id": "b46", "title": "The character of the Indian Kharos \u0323t \u0323h\u012b script and the \"Sanskrit Revolution\": A writing system between identity and assimilation", "authors": [ { "first": "Ingo", "middle": [], "last": "Strauch", "suffix": "" } ], "year": 2012, "venue": "The Idea of Writing: Writing Across Borders", "volume": "", "issue": "", "pages": "131--168", "other_ids": { "DOI": [ "10.1163/9789004217003_008" ] }, "num": null, "urls": [], "raw_text": "Ingo Strauch. 2012. The character of the Indian Kharos \u0323t \u0323h\u012b script and the \"Sanskrit Revolution\": A writing system between identity and assimilation. In Alexander J. de Voogt and Joachim Friedrich Quack, editors, The Idea of Writing: Writing Across Borders, pages 131-168. Brill, Leiden\u037e Boston.", "links": null }, "BIBREF47": { "ref_id": "b47", "title": "Thrax: An open source grammar compiler built on OpenFst", "authors": [ { "first": "Terry", "middle": [], "last": "Tai", "suffix": "" }, { "first": "Wojciech", "middle": [], "last": "Skut", "suffix": "" }, { "first": "Richard", "middle": [], "last": "Sproat", "suffix": "" } ], "year": 2011, "venue": "Proc. of IEEE Automatic Speech Recog nition and Understanding Workshop (ASRU)", "volume": "12", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "Terry Tai, Wojciech Skut, and Richard Sproat. 2011. Thrax: An open source grammar compiler built on OpenFst. In Proc. of IEEE Automatic Speech Recog nition and Understanding Workshop (ASRU), vol ume 12, Hawaii, USA.", "links": null }, "BIBREF48": { "ref_id": "b48", "title": "The Unicode Stan dard", "authors": [ { "first": "Unicode", "middle": [], "last": "Consortium", "suffix": "" } ], "year": 2019, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "Unicode Consortium. 2019. The Unicode Stan dard. Online: http://www.unicode.org/versions/ Unicode12.1.0/. Version 12.1.0, Mountain View, CA.", "links": null }, "BIBREF49": { "ref_id": "b49", "title": "The Conversion of Scripts: Its Nature, History, and Utilization. Information sci ences series", "authors": [ { "first": "H", "middle": [], "last": "Hans", "suffix": "" }, { "first": "", "middle": [], "last": "Wellisch", "suffix": "" } ], "year": 1978, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "Hans H. Wellisch. 1978. The Conversion of Scripts: Its Nature, History, and Utilization. Information sci ences series. John Wiley & Sons, New York.", "links": null }, "BIBREF50": { "ref_id": "b50", "title": "Regular languages", "authors": [ { "first": "Sheng", "middle": [], "last": "Yu", "suffix": "" } ], "year": 1997, "venue": "Grzegorz Rozenberg and Arto Salomaa", "volume": "1", "issue": "", "pages": "41--110", "other_ids": { "DOI": [ "10.1007/978-3-642-59136-5" ] }, "num": null, "urls": [], "raw_text": "Sheng Yu. 1997. Regular languages. In Grzegorz Rozenberg and Arto Salomaa, editors, Handbook of Formal Languages, volume 1: Word, Language, Grammar, pages 41-110. Springer, Berlin.", "links": null }, "BIBREF51": { "ref_id": "b51", "title": "Neural models of text normal ization for speech applications", "authors": [ { "first": "Hao", "middle": [], "last": "Zhang", "suffix": "" }, { "first": "Richard", "middle": [], "last": "Sproat", "suffix": "" }, { "first": "Axel", "middle": [ "H" ], "last": "Ng", "suffix": "" }, { "first": "Felix", "middle": [], "last": "Stahlberg", "suffix": "" }, { "first": "Xiaochang", "middle": [], "last": "Peng", "suffix": "" }, { "first": "Kyle", "middle": [], "last": "Gorman", "suffix": "" }, { "first": "Brian", "middle": [], "last": "Roark", "suffix": "" } ], "year": 2019, "venue": "Computational Lin guistics", "volume": "45", "issue": "2", "pages": "293--337", "other_ids": { "DOI": [ "10.1162/COLI_a_00349" ] }, "num": null, "urls": [], "raw_text": "Hao Zhang, Richard Sproat, Axel H Ng, Felix Stahlberg, Xiaochang Peng, Kyle Gorman, and Brian Roark. 2019. Neural models of text normal ization for speech applications. Computational Lin guistics, 45(2):293-337.", "links": null }, "BIBREF52": { "ref_id": "b52", "title": "The NiuTrans machine translation systems for WMT20", "authors": [ { "first": "Yuhao", "middle": [], "last": "Zhang", "suffix": "" }, { "first": "Ziyang", "middle": [], "last": "Wang", "suffix": "" }, { "first": "Runzhe", "middle": [], "last": "Cao", "suffix": "" }, { "first": "Binghao", "middle": [], "last": "Wei", "suffix": "" }, { "first": "Weiqiao", "middle": [], "last": "Shan", "suffix": "" }, { "first": "Shuhan", "middle": [], "last": "Zhou", "suffix": "" }, { "first": "Abudurexiti", "middle": [], "last": "Re Heman", "suffix": "" }, { "first": "Tao", "middle": [], "last": "Zhou", "suffix": "" }, { "first": "Xin", "middle": [], "last": "Zeng", "suffix": "" }, { "first": "Laohu", "middle": [], "last": "Wang", "suffix": "" }, { "first": "Yongyu", "middle": [], "last": "Mu", "suffix": "" }, { "first": "Jingnan", "middle": [], "last": "Zhang", "suffix": "" }, { "first": "Xiaoqian", "middle": [], "last": "Liu", "suffix": "" }, { "first": "Xuanjun", "middle": [], "last": "Zhou", "suffix": "" }, { "first": "Yinqiao", "middle": [], "last": "Li", "suffix": "" }, { "first": "Bei", "middle": [], "last": "Li", "suffix": "" }, { "first": "Tong", "middle": [], "last": "Xiao", "suffix": "" }, { "first": "Jingbo", "middle": [], "last": "Zhu", "suffix": "" } ], "year": 2020, "venue": "Proceedings of the Fifth Conference on Machine Translation", "volume": "", "issue": "", "pages": "338--345", "other_ids": {}, "num": null, "urls": [], "raw_text": "Yuhao Zhang, Ziyang Wang, Runzhe Cao, Binghao Wei, Weiqiao Shan, Shuhan Zhou, Abudurexiti Re heman, Tao Zhou, Xin Zeng, Laohu Wang, Yongyu Mu, Jingnan Zhang, Xiaoqian Liu, Xuanjun Zhou, Yinqiao Li, Bei Li, Tong Xiao, and Jingbo Zhu. 2020. The NiuTrans machine translation systems for WMT20. In Proceedings of the Fifth Conference on Machine Translation, pages 338-345, Online. Asso ciation for Computational Linguistics.", "links": null } }, "ref_entries": { "FIGREF0": { "text": "String acceptors for Gujarati word \u0aa6\u0ab8 (\u27e8dasa\u27e9, \"ten\") over an alphabet of Unicode code points (top) and bytes (bottom).", "uris": null, "num": null, "type_str": "figure" }, "FIGREF1": { "text": "Simplified construction of the wellformed automaton .", "uris": null, "num": null, "type_str": "figure" }, "FIGREF2": { "text": "Romanization of Sinhala words \u0d91\u0d9a (\"one\") and \u0dd9\u0daf\u0d9a (\"two\") into \u27e8eka\u27e9 and \u27e8deka\u27e9, respectively.", "uris": null, "num": null, "type_str": "figure" }, "FIGREF3": { "text": "Simplified construction of the transliteration transducer \u2110.", "uris": null, "num": null, "type_str": "figure" }, "FIGREF4": { "text": "Software dependency diagrams for the three modes of operation: compile stage (left), Python run time (center) and C++ runtime (right).", "uris": null, "num": null, "type_str": "figure" }, "FIGREF5": { "text": "Runtime Python interface example.", "uris": null, "num": null, "type_str": "figure" }, "TABREF0": { "content": "
NameIdIV DVc cO
BengaliBENG 16 13 435
Devanagari DEVA 19 17 454
GujaratiGUJR 16 15 395
GurmukhiGURU 129 398
KannadaKNDA 17 15 393
Malayalam MLYM 16 16 38 10
OriyaORYA 14 13 385
SinhalaSINH 18 17 412
TamilTAML 12 11 271
TeluguTELU 16 15 385
", "text": "The scripts of interest have evolved from the an cient Br\u0101hm\u012b writing system that was recorded", "num": null, "html": null, "type_str": "table" }, "TABREF1": { "content": "
: Sizes of core graphemic classes: Independent
vowels (IV), dependent vowel diacritics (DV), conso
nants (c), coda symbols (cO).
", "text": "", "num": null, "html": null, "type_str": "table" }, "TABREF3": { "content": "", "text": "NFC examples for Devanagari.", "num": null, "html": null, "type_str": "table" }, "TABREF5": { "content": "
", "text": "Examples for additions to ISO 15919.", "num": null, "html": null, "type_str": "table" }, "TABREF9": { "content": "
: Properties of script FSTs arranged by operation and symbol types (Unicode code points and UTF8 bytes),
where \u2110 denotes the ISO transliteration operation,is the NFC normalization, denotes visual normalization,
andis the wellformed check. The numbers of states and arcs are denoted byand, respectively.
BrahmicBrahmicBrahmic
Offline: CompileRuntime: PythonRuntime: C++
PyniniThraxPyniniThrax
OpenFstOpenFstOpenFst
", "text": "", "num": null, "html": null, "type_str": "table" }, "TABREF10": { "content": "
3cd nisaba
4# Compile the transducers and tests.
8# Compile Thrax rewrite helper tool.
9bazel build -c opt @org_opengrm_thrax//:rewrite-tester
10# Run visual normalization for Kannada.
11bazel-bin/external/org_opengrm_thrax/rewrite-tester \\
12--far=bazel-bin/nisaba/brahmic/visual_norm.far \\
13--rules=KNDA < words.txt
Figure 6: Compiling the transducers.
import unittest
from nisaba import brahmic
class BrahmicTest(unittest.TestCase):
def testBasicOperations(self):
# Check romanization.
iso_to_deva = brahmic.IsoTo('Deva')
self.assertEqual('\u0915\u094d \u093c \u0932\u092c',
", "text": "git clone https://github.com/google-research/nisaba.git bazel build -c opt //nisaba/brahmic/... bazel test -c opt //nisaba/brahmic/...", "num": null, "html": null, "type_str": "table" }, "TABREF12": { "content": "", "text": "Percentage of types and tokens changed by vi sual normalization from native script Wikipedia train ing partitions of the Dakshina dataset.", "num": null, "html": null, "type_str": "table" } } } }