lguenth/backsum · Datasets at Hugging Face

source_text stringlengths 27 368	label int64 0 1	target_text stringlengths 1 5.38k
In this work, they take a more direct approach and treat a word type and its allowed POS tags as a primary element of the model.	0	(2010) and the posterior regular- ization HMM of GracÂ¸a et al.
This architecture provides a uniform framework in which it is easy to incorporate not only listed dictionary entries but also morphological derivatives, and models for personal names and foreign names in transliteration.	0	More formally, we start by representing the dictionary D as a Weighted Finite State TransÂ ducer (WFST) (Pereira, Riley, and Sproat 1994).
Other kinds of productive word classes, such as company names, abbreviations,and place names can easily be handled given appropriate models.	0	In the denomi 11 We have two such lists, one containing about 17,000 full names, and another containing frequencies of.
This paper conducted research in the area of automatic paraphrase discovery.	0	The work reported here is closely related to [Ha- segawa et al. 04].
This paper talks about KenLM: Faster and Smaller Language Model Queries.	0	Throughout this paper we compare with several packages: SRILM 1.5.12 (Stolcke, 2002) is a popular toolkit based on tries used in several decoders.
They have made use of local and global features to deal with the instances of same token in a document.	0	Borth 2MUC data can be obtained from the Linguistic Data Consortium: http://www.ldc.upenn.edu 3Training data for IdentiFinder is actually given in words (i.e., 650K & 790K words), rather than tokens Table 6: Comparison of results for MUC7 wick (1999) reported using dictionaries of person first names, corporate names and suffixes, colleges and universities, dates and times, state abbreviations, and world regions.
The TRIE model uses less memory than the smallest lossless alternative and is still faster than SRILM.	0	IRSTLM (Federico et al., 2008) is an open-source toolkit for building and querying language models.
They found replacing it with a ranked evaluation to be more suitable.	0	The judgements tend to be done more in form of a ranking of the different systems.
We chose one of four labels for each example: location, person, organization, or noise where the noise category was used for items that were outside the three categories.	0	(8) can now be rewritten5 as which is of the same form as the function Zt used in AdaBoost.
Evaluation was done automatically using the BLEU score and manually on fluency and adequacy.	1	We asked participants to each judge 200–300 sentences in terms of fluency and adequacy, the most commonly used manual evaluation metrics.
They proposed an unsupervised method to discover paraphrases from a large untagged corpus.	0	Sometimes extracted phrases by themselves are not meaningful to consider without context, but we set the following criteria.
Nevertheless, only a part of this corpus (10 texts), which the authors name "core corpus", is annotated with all this information.	0	In an experiment on automatic rhetorical parsing, the RST-annotations and PoS tags were used by (Reitter 2003) as a training corpus for statistical classification with Support Vector Machines.
The first method builds on results from (Yarowsky 95) and (Blum and Mitchell 98).	0	Assume that the two classifiers are "rote learners": that is, 1.1 and 12 are defined through look-up tables that list a label for each member of X1 or X2.
Other kinds of productive word classes, such as company names, abbreviations,and place names can easily be handled given appropriate models.	0	This is orthographically represented as 7C.
It is probably the first analysis of Arabic parsing of this kind.	0	Cohen and Smith (2007) chose a metric like SParseval (Roark et al., 2006) that first aligns the trees and then penalizes segmentation errors with an edit-distance metric.
They showed the efficacy of graph-based label propagation for projecting part-of-speech information across languages.	0	To bridge this gap, we consider a practically motivated scenario, in which we want to leverage existing resources from a resource-rich language (like English) when building tools for resource-poor foreign languages.1 We assume that absolutely no labeled training data is available for the foreign language of interest, but that we have access to parallel data with a resource-rich language.
They proposed a single joint model for performing both morphological segmentation and syntactic disambiguation which bypasses the associated circularity.	0	We further report SYNCS, the parsing metric of Cohen and Smith (2007), to facilitate the comparison.
It is probably the first analysis of Arabic parsing of this kind.	0	97 78.
There are clustering approaches that assign a single POS tag to each word type.	0	Mo del Hy per par am . E n g li s h1 1 m-1 D a n i s h1 1 m-1 D u t c h1 1 m-1 G er m a n1 1 m-1 Por tug ues e1 1 m-1 S p a ni s h1 1 m-1 S w e di s h1 1 m-1 1T W be st me dia n 45.
Most IE researchers have been creating paraphrase knowledge by hand and specific tasks.	0	At this step, we will try to link those sets, and put them into a single cluster.
Throughout this paper we compare with several packages: SRILM 1.5.12 (Stolcke, 2002) is a popular toolkit based on tries used in several decoders.	0	Therefore, performance is more closely tied to the underlying data structure than to the cache.
Bean and Riloff also used bootstrapping to extend their semantic compatibility model, proposed using caseframe network for anaphora resolution, information extraction patterns to identify contextual clues for determining compatibility between NPs.	0	Ex: He was found in San Jose, where ...
Due to many similarly performing systems, the author was not able to draw strong conclusions on the question of correlation of manual and automatic evaluation metrics.	0	Not every annotator was fluent in both the source and the target language.
The manual evaluation of scoring translation on a graded scale from 1â5 seems to be very hard to perform.	0	However, ince we extracted the test corpus automatically from web sources, the reference translation was not always accurate — due to sentence alignment errors, or because translators did not adhere to a strict sentence-by-sentence translation (say, using pronouns when referring to entities mentioned in the previous sentence).
The TRIE model uses less memory than the smallest lossless alternative and is still faster than SRILM.	0	RandLM and SRILM also remove context that will not extend, but SRILM performs a second lookup in its trie whereas our approach has minimal additional cost.
Other kinds of productive word classes, such as company names, abbreviations,and place names can easily be handled given appropriate models.	0	However, this result is consistent with the results of exÂ periments discussed in Wu and Fung (1994).
They have made use of local and global features to deal with the instances of same token in a document.	0	We often encounter sentences that are highly ambiguous in themselves, without some prior knowledge of the entities concerned.
In this paper the author evaluates machine translation performance for six European language pairs that participated in a shared task: translating French, German, Spanish texts to English and back.	0	While we had up to 11 submissions for a translation direction, we did decide against presenting all 11 system outputs to the human judge.
The authors cluster NE instance pairs based on the words in the context using bag-of-words methods.	0	(If the TF/IDF score of that word is below a threshold, the phrase is discarded.)
Most IE researchers have been creating paraphrase knowledge by hand and specific tasks.	0	âAgreeâ is a subject control verb, which dominates another verb whose subject is the same as that of âagreeâ; the latter verb is generally the one of interest for extraction.
It is probably the first analysis of Arabic parsing of this kind.	0	3.2 Inter-annotator Agreement.
The manual evaluation of scoring translation on a graded scale from 1â5 seems to be very hard to perform.	0	More judgements would have enabled us to make better distinctions, but it is not clear what the upper limit is.
This paper presents methods to query N-gram language models, minimizing time and space costs.	0	As noted for the perplexity task, we do not expect cache to grow substantially with model size, so RandLM remains a low-memory option.
For the perplexity and translation tasks, we used SRILM to build a 5-gram English language model on 834 million tokens from Europarl v6 (Koehn, 2005) and the 2011 Workshop on Machine Translation News Crawl corpus with duplicate lines removed.	0	The index in this array is the vocabulary identifier.
The code is open source, has minimal dependencies, and offers both C++ and Java interfaces for integration.	0	Hash tables are a common sparse mapping technique used by SRILM’s default and BerkeleyLM’s hashed variant.
They found replacing it with a ranked evaluation to be more suitable.	0	Many human evaluation metrics have been proposed.
Across eight European languages, their approach results in an average absolute improvement of 10.4% over a state-of-the-art baseline, and 16.7% over vanilla hidden Markov models induced with the Expectation Maximization algorithm.	0	It can be shown that this objective is convex in q.
The model incorporates various recent techniques for incorporating and manipulating linguistic knowledge using finite-state transducers.	0	However, they list two sets, one consisting of 28 fragments and the other of 22 fragments, in which they had 0% recall and precision.
Their work is closely related to recent approaches that incorporate the sparsity constraint into the POS induction process.	0	In contrast to these approaches, our method directly incorporates these constraints into the structure of the model.
In this paper the author evaluates machine translation performance for six European language pairs that participated in a shared task: translating French, German, Spanish texts to English and back.	0	Manual and Automatic Evaluation of Machine Translation between European Languages
It is probably the first analysis of Arabic parsing of this kind.	0	For example, we might have VP â VB NP PP, where the NP is the subject.
NER is useful in many NLP applications such as information extraction, question answering, etc .Chieu and Ng have shown that the maximum entropy framework is able to use global information directly from various sources.	0	If any of the tokens from to is in Person-Prefix- List, then another feature Person-Prefix is set to 1.
Their results show that their high performance NER use less training data than other systems.	0	By using the output of a hand-coded system such as Proteus, MENE can improve its performance, and can even outperform IdentiFinder (Borthwick, 1999).
They focused on phrases which two Named Entities, and proceed in two stages.	0	As we shall see, most of the linked sets are paraphrases.
This paper discusses the Potsdam Commentary Corpus, a corpus of german assembeled by potsdam university.	0	This fact annoyed especially his dog...).
It is well-known that English constituency parsing models do not generalize to other languages and treebanks.	0	We report micro-averaged (whole corpus) and macro-averaged (per sentence) scores along add a constraint on the removal of punctuation, which has a single tag (PUNC) in the ATB.
The authors show that PATB is similar to other tree-banks but that annotation consistency remains low.	0	pre-processing.
The authors in this paper describe a search procedure for statistical machine translation (MT) based on dynamic programming (DP).	0	The computing time is given in terms of CPU time per sentence (on a 450MHz PentiumIIIPC).
There is no global pruning.	0	What is important and is not expressed by the notation is the so-called coverage constraint: each source position j should be 'hit' exactly once by the path of the inverted alignment bI 1 = b1:::bi:::bI . Using the inverted alignments in the maximum approximation, we obtain as search criterion: max I (p(JjI) max eI 1 ( I Yi=1 p(eijeiô1 iô2) max bI 1 I Yi=1 [p(bijbiô1; I; J) p(fbi jei)])) = = max I (p(JjI) max eI 1;bI 1 ( I Yi=1 p(eijeiô1 iô2) p(bijbiô1; I; J) p(fbi jei)])); where the two products over i have been merged into a single product over i. p(eijeiô1 iô2) is the trigram language model probability.
The texts were annotated with the RSTtool.	0	Cur In order to ground such approaches in linguistic observation and description, a multi-level anno 10 For an exposition of the idea as applied to the task of text planning, see (Chiarcos, Stede 2004).
Human judges also pointed out difficulties with the evaluation of long sentences.	0	A few pointed out that adequacy should be broken up into two criteria: (a) are all source words covered?
The authors use linguistic and annotation insights to develop a manually annotated grammar and evaluate it and finally provide a realistic evaluation in which segmentation is performed in a pipeline jointly with parsing.	0	A cell in the bottom row of the parse chart is required for each potential whitespace boundary.
The approach has been successfully tested on the 8 000-word Verbmobil task.	0	The proof is given in (Tillmann, 2000).
However, these approaches are ill-equipped to directly represent type-based constraints such as sparsity.	0	our full model yields 39.3% average error reduction across languages when compared to the basic configuration (1TW).
In order to create good-sized vectors for similarity calculation, they had to set a high frequency threshold.	0	A total of 13,976 phrases were grouped.
They proposed an unsupervised method to discover paraphrases from a large untagged corpus.	0	The number of NE instance pairs used in their experiment is less than half of our method.
In order to handle the necessary word reordering as an optimization problem within the dynamic programming approach, they describe a solution to the traveling salesman problem (TSP) which is based on dynamic programming.	0	In Section 4, we present the performance measures used and give translation results on the Verbmobil task.
This paper talks about KenLM: Faster and Smaller Language Model Queries.	0	This information is readily available in TRIE where adjacent records with equal pointers indicate no further extension of context is possible.
This architecture provides a uniform framework in which it is easy to incorporate not only listed dictionary entries but also morphological derivatives, and models for personal names and foreign names in transliteration.	0	However, there is again local grammatical information that should favor the split in the case of (1a): both .ma3 'horse' and .ma3 lu4 are nouns, but only .ma3 is consistent with the classifier pil, the classifier for horses.21 By a similar argument, the preference for not splitting , lm could be strengthened in (lb) by the observation that the classifier 1'1* tiao2 is consistent with long or winding objects like , lm ma3lu4 'road' but not with,ma3 'horse.'
They used graph-based label propagation for cross-lingual knowledge transfer and used the projected labels as features in an unsupervised model.	0	We adopted this state-of-the-art model because it makes it easy to experiment with various ways of incorporating our novel constraint feature into the log-linear emission model.
These clusters are computed using an SVD variant without relying on transitional structure.	0	We observe similar trends when using another measure â type-level accuracy (defined as the fraction of words correctly assigned their majority tag), according to which La ng ua ge M etr ic B K 10 E M B K 10 L B F G S G 10 F EA T S B es t F EA T S M ed ia n E ng lis h 1 1 m 1 4 8 . 3 6 8 . 1 5 6 . 0 7 5 . 5 â â 5 0 . 9 6 6 . 4 4 7 . 8 6 6 . 4 D an is h 1 1 m 1 4 2 . 3 6 6 . 7 4 2 . 6 5 8 . 0 â â 5 2 . 1 6 1 . 2 4 3 . 2 6 0 . 7 D ut ch 1 1 m 1 5 3 . 7 6 7 . 0 5 5 . 1 6 4 . 7 â â 5 6 . 4 6 9 . 0 5 1 . 5 6 7 . 3 Po rtu gu es e 1 1 m 1 5 0 . 8 7 5 . 3 4 3 . 2 7 4 . 8 44 .5 69 .2 6 4 . 1 7 4 . 5 5 6 . 5 7 0 . 1 S pa ni sh 1 1 m 1 â â 4 0 . 6 7 3 . 2 â â 5 8 . 3 6 8 . 9 5 0 . 0 5 7 . 2 Table 4: Comparison of our method (FEATS) to state-of-the-art methods.
They used graph-based label propagation for cross-lingual knowledge transfer and used the projected labels as features in an unsupervised model.	0	The graph was constructed using 2 million trigrams; we chose these by truncating the parallel datasets up to the number of sentence pairs that contained 2 million trigrams.
This paper conducted research in the area of automatic paraphrase discovery.	0	When a company buys another company, a paying event can occur, but these two phrases do not indicate the same event.
The code is open source, has minimal dependencies, and offers both C++ and Java interfaces for integration.	0	The size of TRIE is particularly sensitive to F1092 c11, so vocabulary filtering is quite effective at reducing model size.
The authors cluster NE instance pairs based on the words in the context using bag-of-words methods.	0	Automatic Paraphrase Discovery based on Context and Keywords between NE Pairs
Evaluation was done automatically using the BLEU score and manually on fluency and adequacy.	0	Often, two systems can not be distinguished with a confidence of over 95%, so there are ranked the same.
Human judges also pointed out difficulties with the evaluation of long sentences.	0	Since different judges judged different systems (recall that judges were excluded to judge system output from their own institution), we normalized the scores.
These clusters are computed using an SVD variant without relying on transitional structure.	0	In total there are O(K 2) parameters associated with the transition parameters.
It also incorporates the Good-Turing methodin estimating the likelihoods of previously unseen constructions, including morphological derivatives and personal names.	0	paper, and is missing 6 examples from the A set.
The PROBING data structure uses linear probing hash tables and is designed for speed.	0	IRSTLM 5.60.02 (Federico et al., 2008) is a sorted trie implementation designed for lower memory consumption.
Human judges also pointed out difficulties with the evaluation of long sentences.	0	The confidence intervals are computed by bootstrap resampling for BLEU, and by standard significance testing for the manual scores, as described earlier in the paper.
Das and Petrov, in this paper, approached inducing unsupervised part-of-speech taggers for languages that had no labeled training data, but had translated text in a resource-rich language.	0	Supervised learning approaches have advanced the state-of-the-art on a variety of tasks in natural language processing, resulting in highly accurate systems.
Using a treebank grammar, the data-driven lexicon, and a linguistically motivated unknown-tokens handling technique their model outperformed previous pipelined, integrated or factorized systems for Hebrew morphological and syntactic processing, yielding an error reduction of 12% over the best published results so far.	0	Tsarfaty (2006) used a morphological analyzer (Segal, 2000), a PoS tagger (Bar-Haim et al., 2005), and a general purpose parser (Schmid, 2000) in an integrated framework in which morphological and syntactic components interact to share information, leading to improved performance on the joint task.
From a DP-based solution to the traveling salesman problem, they present a novel technique to restrict the possible word reordering between source and target language in order to achieve an eÃcient search algorithm.	0	For the experiments, we use a simple preprocessing step.
They found replacing it with a ranked evaluation to be more suitable.	0	The easiest language pair according to BLEU (English-French: 28.33) received worse manual scores than the hardest (English-German: 14.01).
They extended previous work on discriminative weighting by using a finer granularity, focusing on the properties of instances rather than corpus components, and used simpler training procedure.	0	Finally, we make some improvements to baseline approaches.
they evaluate various specific aspects of the segmentation, as well as the overall segmentation performance.	0	In Table 5 we present results from small test corÂ pora for the productive affixes handled by the current version of the system; as with names, the segmentation of morphologically derived words is generally either right or wrong.
In order to handle the necessary word reordering as an optimization problem within the dynamic programming approach, they describe a solution to the traveling salesman problem (TSP) which is based on dynamic programming.	0	Two subjects are each given a calendar and they are asked to schedule a meeting.
This paper talks about Unsupervised Models for Named Entity Classification.	0	(6), with W+ > W_.
Using less training data than other systems, their NER can perform as well as other state-of-the-art NERs.	0	Note that we check for , the word preceding the consecutive sequence of initCaps tokens, since person prefixes like Mr., Dr., etc are not part of person names, whereas corporate suffixes like Corp., Inc., etc are part of corporate names.
Other kinds of productive word classes, such as company names, abbreviations,and place names can easily be handled given appropriate models.	0	Word type N % Dic tion ary entr ies 2 , 5 4 3 9 7 . 4 7 Mor pho logi call y deri ved wor ds 3 0 . 1 1 Fore ign tran slite rati ons 9 0 . 3 4 Per son al na mes 5 4 2 . 0 7 cases.
The PROBING data structure uses linear probing hash tables and is designed for speed.	0	We implement two data structures: PROBING, designed for speed, and TRIE, optimized for memory.
they evaluate various specific aspects of the segmentation, as well as the overall segmentation performance.	0	For derived words that occur in our corpus we can estimate these costs as we would the costs for an underived dictionary entry.
The use of global features has shown excellent result in the performance on MUC-6 and MUC-7 test data.	0	The features we used can be divided into 2 classes: local and global.
The authors believe automatic paraphrase discovery is an important component for building a fully automatic information extraction system.	0	Find keywords for each NE pair When we look at the contexts for each domain, we noticed that there is one or a few important words which indicate the relation between the NEs (for example, the word âunitâ for the phrase âa unit ofâ).
In this paper, Ben and Riloff present a coreference resolver called BABAR that focuses on the use of contextual-role knowledge for coreference resolution.	0	For example, if {N P1, N P2, N P3} are all coreferent, then each NP must be linked to one of the other two NPs.
The departure from the traditional token-based tagging approach allow them to explicitly capture type-level distributional properties of valid POS tag assignments as part of the model.	0	Once the lexicon has been drawn, the model proceeds similarly to the standard token-level HMM: Emission parameters Î¸ are generated conditioned on tag assignments T . We also draw transition parameters Ï.
The AdaBoost algorithm was developed for supervised learning.	0	2.
The corpus was annoted with different linguitic information.	0	Clearly this poses a number of research challenges, though, such as the applicability of tag sets across different languages.
This paper talks about KenLM: Faster and Smaller Language Model Queries.	0	This makes memory usage comparable to our PROBING model.
The manual evaluation of scoring translation on a graded scale from 1â5 seems to be very hard to perform.	0	This work was supported in part under the GALE program of the Defense Advanced Research Projects Agency, Contract No.
Finally, several coreference systems have successfully incorporated anaphoricity determination modules.	0	Table 1: Syntactic Seeding Heuristics BABARâs reliable case resolution heuristics produced a substantial set of anaphor/antecedent resolutions that will be the training data used to learn contextual role knowledge.
This paper offers a broad insight into of Arabic constituency parsing by analyzing the interplay of linguistic phenomena, annotation choices, and model design.	0	The ATB annotation guidelines specify that proper nouns should be specified with a flat NP (a).
This paper offers a broad insight into of Arabic constituency parsing by analyzing the interplay of linguistic phenomena, annotation choices, and model design.	0	Also surprising is the low test set OOV rate given the possibility of morphological variation in Arabic.
The PROBING data structure uses linear probing hash tables and is designed for speed.	0	Several methods exist to handle collisions; we use linear probing because it has less memory overhead when entries are small.
In order to create good-sized vectors for similarity calculation, they had to set a high frequency threshold.	0	This overview is illustrated in Figure 1.
they evaluate various specific aspects of the segmentation, as well as the overall segmentation performance.	0	Mandarin exhibits several such processes, including A-not-A question formation, ilÂ lustrated in (3a), and adverbial reduplication, illustrated in (3b): 3.
Finally, several coreference systems have successfully incorporated anaphoricity determination modules.	0	5 Related Work.
Bean and Riloff also used bootstrapping to extend their semantic compatibility model, proposed using caseframe network for anaphora resolution, information extraction patterns to identify contextual clues for determining compatibility between NPs.	0	Table 1 briefly describes the seven syntactic heuristics used by BABAR to resolve noun phrases.
The authors use linguistic and annotation insights to develop a manually annotated grammar and evaluate it and finally provide a realistic evaluation in which segmentation is performed in a pipeline jointly with parsing.	0	Instead, we extend the variation n-gram method of Dickinson (2005) to compare annotation error rates in the WSJ and ATB.

In this work, they take a more direct approach and treat a word type and its allowed POS tags as a primary element of the model.

0

(2010) and the posterior regular- ization HMM of GracÂ¸a et al.

This architecture provides a uniform framework in which it is easy to incorporate not only listed dictionary entries but also morphological derivatives, and models for personal names and foreign names in transliteration.

0

More formally, we start by representing the dictionary D as a Weighted Finite State TransÂ ducer (WFST) (Pereira, Riley, and Sproat 1994).

Other kinds of productive word classes, such as company names, abbreviations,and place names can easily be handled given appropriate models.

0

In the denomi 11 We have two such lists, one containing about 17,000 full names, and another containing frequencies of.

This paper conducted research in the area of automatic paraphrase discovery.

0

The work reported here is closely related to [Ha- segawa et al. 04].

This paper talks about KenLM: Faster and Smaller Language Model Queries.

0

Throughout this paper we compare with several packages: SRILM 1.5.12 (Stolcke, 2002) is a popular toolkit based on tries used in several decoders.

They have made use of local and global features to deal with the instances of same token in a document.

0

Borth 2MUC data can be obtained from the Linguistic Data Consortium: http://www.ldc.upenn.edu 3Training data for IdentiFinder is actually given in words (i.e., 650K & 790K words), rather than tokens Table 6: Comparison of results for MUC7 wick (1999) reported using dictionaries of person first names, corporate names and suffixes, colleges and universities, dates and times, state abbreviations, and world regions.

The TRIE model uses less memory than the smallest lossless alternative and is still faster than SRILM.

0

IRSTLM (Federico et al., 2008) is an open-source toolkit for building and querying language models.

They found replacing it with a ranked evaluation to be more suitable.

0

The judgements tend to be done more in form of a ranking of the different systems.

We chose one of four labels for each example: location, person, organization, or noise where the noise category was used for items that were outside the three categories.

0

(8) can now be rewritten5 as which is of the same form as the function Zt used in AdaBoost.

Evaluation was done automatically using the BLEU score and manually on fluency and adequacy.

1

We asked participants to each judge 200–300 sentences in terms of fluency and adequacy, the most commonly used manual evaluation metrics.

They proposed an unsupervised method to discover paraphrases from a large untagged corpus.

0

Sometimes extracted phrases by themselves are not meaningful to consider without context, but we set the following criteria.

Nevertheless, only a part of this corpus (10 texts), which the authors name "core corpus", is annotated with all this information.

0

In an experiment on automatic rhetorical parsing, the RST-annotations and PoS tags were used by (Reitter 2003) as a training corpus for statistical classification with Support Vector Machines.

The first method builds on results from (Yarowsky 95) and (Blum and Mitchell 98).

0

Assume that the two classifiers are "rote learners": that is, 1.1 and 12 are defined through look-up tables that list a label for each member of X1 or X2.

Other kinds of productive word classes, such as company names, abbreviations,and place names can easily be handled given appropriate models.

0

This is orthographically represented as 7C.

It is probably the first analysis of Arabic parsing of this kind.

0

Cohen and Smith (2007) chose a metric like SParseval (Roark et al., 2006) that first aligns the trees and then penalizes segmentation errors with an edit-distance metric.

They showed the efficacy of graph-based label propagation for projecting part-of-speech information across languages.

0

To bridge this gap, we consider a practically motivated scenario, in which we want to leverage existing resources from a resource-rich language (like English) when building tools for resource-poor foreign languages.1 We assume that absolutely no labeled training data is available for the foreign language of interest, but that we have access to parallel data with a resource-rich language.

They proposed a single joint model for performing both morphological segmentation and syntactic disambiguation which bypasses the associated circularity.

0

We further report SYNCS, the parsing metric of Cohen and Smith (2007), to facilitate the comparison.

It is probably the first analysis of Arabic parsing of this kind.

0

97 78.

There are clustering approaches that assign a single POS tag to each word type.

0

Mo del Hy per par am . E n g li s h1 1 m-1 D a n i s h1 1 m-1 D u t c h1 1 m-1 G er m a n1 1 m-1 Por tug ues e1 1 m-1 S p a ni s h1 1 m-1 S w e di s h1 1 m-1 1T W be st me dia n 45.

Most IE researchers have been creating paraphrase knowledge by hand and specific tasks.

0

At this step, we will try to link those sets, and put them into a single cluster.

Throughout this paper we compare with several packages: SRILM 1.5.12 (Stolcke, 2002) is a popular toolkit based on tries used in several decoders.

0

Therefore, performance is more closely tied to the underlying data structure than to the cache.

Bean and Riloff also used bootstrapping to extend their semantic compatibility model, proposed using caseframe network for anaphora resolution, information extraction patterns to identify contextual clues for determining compatibility between NPs.

0

Ex: He was found in San Jose, where ...

Due to many similarly performing systems, the author was not able to draw strong conclusions on the question of correlation of manual and automatic evaluation metrics.

0

Not every annotator was fluent in both the source and the target language.

The manual evaluation of scoring translation on a graded scale from 1â5 seems to be very hard to perform.

0

However, ince we extracted the test corpus automatically from web sources, the reference translation was not always accurate — due to sentence alignment errors, or because translators did not adhere to a strict sentence-by-sentence translation (say, using pronouns when referring to entities mentioned in the previous sentence).

The TRIE model uses less memory than the smallest lossless alternative and is still faster than SRILM.

0

RandLM and SRILM also remove context that will not extend, but SRILM performs a second lookup in its trie whereas our approach has minimal additional cost.

Other kinds of productive word classes, such as company names, abbreviations,and place names can easily be handled given appropriate models.

0

However, this result is consistent with the results of exÂ periments discussed in Wu and Fung (1994).

They have made use of local and global features to deal with the instances of same token in a document.

0

We often encounter sentences that are highly ambiguous in themselves, without some prior knowledge of the entities concerned.

In this paper the author evaluates machine translation performance for six European language pairs that participated in a shared task: translating French, German, Spanish texts to English and back.

0

While we had up to 11 submissions for a translation direction, we did decide against presenting all 11 system outputs to the human judge.

The authors cluster NE instance pairs based on the words in the context using bag-of-words methods.

0

(If the TF/IDF score of that word is below a threshold, the phrase is discarded.)

Most IE researchers have been creating paraphrase knowledge by hand and specific tasks.

0

âAgreeâ is a subject control verb, which dominates another verb whose subject is the same as that of âagreeâ; the latter verb is generally the one of interest for extraction.

It is probably the first analysis of Arabic parsing of this kind.

0

3.2 Inter-annotator Agreement.

The manual evaluation of scoring translation on a graded scale from 1â5 seems to be very hard to perform.

0

More judgements would have enabled us to make better distinctions, but it is not clear what the upper limit is.

This paper presents methods to query N-gram language models, minimizing time and space costs.

0

As noted for the perplexity task, we do not expect cache to grow substantially with model size, so RandLM remains a low-memory option.

For the perplexity and translation tasks, we used SRILM to build a 5-gram English language model on 834 million tokens from Europarl v6 (Koehn, 2005) and the 2011 Workshop on Machine Translation News Crawl corpus with duplicate lines removed.

0

The index in this array is the vocabulary identifier.

The code is open source, has minimal dependencies, and offers both C++ and Java interfaces for integration.

0

Hash tables are a common sparse mapping technique used by SRILM’s default and BerkeleyLM’s hashed variant.

They found replacing it with a ranked evaluation to be more suitable.

0

Many human evaluation metrics have been proposed.

Across eight European languages, their approach results in an average absolute improvement of 10.4% over a state-of-the-art baseline, and 16.7% over vanilla hidden Markov models induced with the Expectation Maximization algorithm.

0

It can be shown that this objective is convex in q.

The model incorporates various recent techniques for incorporating and manipulating linguistic knowledge using finite-state transducers.

0

However, they list two sets, one consisting of 28 fragments and the other of 22 fragments, in which they had 0% recall and precision.

Their work is closely related to recent approaches that incorporate the sparsity constraint into the POS induction process.

0

In contrast to these approaches, our method directly incorporates these constraints into the structure of the model.

In this paper the author evaluates machine translation performance for six European language pairs that participated in a shared task: translating French, German, Spanish texts to English and back.

0

Manual and Automatic Evaluation of Machine Translation between European Languages

It is probably the first analysis of Arabic parsing of this kind.

0

For example, we might have VP â VB NP PP, where the NP is the subject.

NER is useful in many NLP applications such as information extraction, question answering, etc .Chieu and Ng have shown that the maximum entropy framework is able to use global information directly from various sources.

0

If any of the tokens from to is in Person-Prefix- List, then another feature Person-Prefix is set to 1.

Their results show that their high performance NER use less training data than other systems.

0

By using the output of a hand-coded system such as Proteus, MENE can improve its performance, and can even outperform IdentiFinder (Borthwick, 1999).

They focused on phrases which two Named Entities, and proceed in two stages.

0

As we shall see, most of the linked sets are paraphrases.

This paper discusses the Potsdam Commentary Corpus, a corpus of german assembeled by potsdam university.

0

This fact annoyed especially his dog...).

It is well-known that English constituency parsing models do not generalize to other languages and treebanks.

0

We report micro-averaged (whole corpus) and macro-averaged (per sentence) scores along add a constraint on the removal of punctuation, which has a single tag (PUNC) in the ATB.

The authors show that PATB is similar to other tree-banks but that annotation consistency remains low.

0

pre-processing.

The authors in this paper describe a search procedure for statistical machine translation (MT) based on dynamic programming (DP).

0

The computing time is given in terms of CPU time per sentence (on a 450MHz PentiumIIIPC).

There is no global pruning.

0

What is important and is not expressed by the notation is the so-called coverage constraint: each source position j should be 'hit' exactly once by the path of the inverted alignment bI 1 = b1:::bi:::bI . Using the inverted alignments in the maximum approximation, we obtain as search criterion: max I (p(JjI) max eI 1 ( I Yi=1 p(eijeiô1 iô2) max bI 1 I Yi=1 [p(bijbiô1; I; J) p(fbi jei)])) = = max I (p(JjI) max eI 1;bI 1 ( I Yi=1 p(eijeiô1 iô2) p(bijbiô1; I; J) p(fbi jei)])); where the two products over i have been merged into a single product over i. p(eijeiô1 iô2) is the trigram language model probability.

The texts were annotated with the RSTtool.

0

Cur In order to ground such approaches in linguistic observation and description, a multi-level anno 10 For an exposition of the idea as applied to the task of text planning, see (Chiarcos, Stede 2004).

Human judges also pointed out difficulties with the evaluation of long sentences.

0

A few pointed out that adequacy should be broken up into two criteria: (a) are all source words covered?

The authors use linguistic and annotation insights to develop a manually annotated grammar and evaluate it and finally provide a realistic evaluation in which segmentation is performed in a pipeline jointly with parsing.

0

A cell in the bottom row of the parse chart is required for each potential whitespace boundary.

The approach has been successfully tested on the 8 000-word Verbmobil task.

0

The proof is given in (Tillmann, 2000).

However, these approaches are ill-equipped to directly represent type-based constraints such as sparsity.

0

our full model yields 39.3% average error reduction across languages when compared to the basic configuration (1TW).

In order to create good-sized vectors for similarity calculation, they had to set a high frequency threshold.

0

A total of 13,976 phrases were grouped.

They proposed an unsupervised method to discover paraphrases from a large untagged corpus.

0

The number of NE instance pairs used in their experiment is less than half of our method.

In order to handle the necessary word reordering as an optimization problem within the dynamic programming approach, they describe a solution to the traveling salesman problem (TSP) which is based on dynamic programming.

0

In Section 4, we present the performance measures used and give translation results on the Verbmobil task.

This paper talks about KenLM: Faster and Smaller Language Model Queries.

0

This information is readily available in TRIE where adjacent records with equal pointers indicate no further extension of context is possible.

This architecture provides a uniform framework in which it is easy to incorporate not only listed dictionary entries but also morphological derivatives, and models for personal names and foreign names in transliteration.

0

However, there is again local grammatical information that should favor the split in the case of (1a): both .ma3 'horse' and .ma3 lu4 are nouns, but only .ma3 is consistent with the classifier pil, the classifier for horses.21 By a similar argument, the preference for not splitting , lm could be strengthened in (lb) by the observation that the classifier 1'1* tiao2 is consistent with long or winding objects like , lm ma3lu4 'road' but not with,ma3 'horse.'

They used graph-based label propagation for cross-lingual knowledge transfer and used the projected labels as features in an unsupervised model.

0

We adopted this state-of-the-art model because it makes it easy to experiment with various ways of incorporating our novel constraint feature into the log-linear emission model.

These clusters are computed using an SVD variant without relying on transitional structure.

0

We observe similar trends when using another measure â type-level accuracy (defined as the fraction of words correctly assigned their majority tag), according to which La ng ua ge M etr ic B K 10 E M B K 10 L B F G S G 10 F EA T S B es t F EA T S M ed ia n E ng lis h 1 1 m 1 4 8 . 3 6 8 . 1 5 6 . 0 7 5 . 5 â â 5 0 . 9 6 6 . 4 4 7 . 8 6 6 . 4 D an is h 1 1 m 1 4 2 . 3 6 6 . 7 4 2 . 6 5 8 . 0 â â 5 2 . 1 6 1 . 2 4 3 . 2 6 0 . 7 D ut ch 1 1 m 1 5 3 . 7 6 7 . 0 5 5 . 1 6 4 . 7 â â 5 6 . 4 6 9 . 0 5 1 . 5 6 7 . 3 Po rtu gu es e 1 1 m 1 5 0 . 8 7 5 . 3 4 3 . 2 7 4 . 8 44 .5 69 .2 6 4 . 1 7 4 . 5 5 6 . 5 7 0 . 1 S pa ni sh 1 1 m 1 â â 4 0 . 6 7 3 . 2 â â 5 8 . 3 6 8 . 9 5 0 . 0 5 7 . 2 Table 4: Comparison of our method (FEATS) to state-of-the-art methods.

They used graph-based label propagation for cross-lingual knowledge transfer and used the projected labels as features in an unsupervised model.

0

The graph was constructed using 2 million trigrams; we chose these by truncating the parallel datasets up to the number of sentence pairs that contained 2 million trigrams.

This paper conducted research in the area of automatic paraphrase discovery.

0

When a company buys another company, a paying event can occur, but these two phrases do not indicate the same event.

The code is open source, has minimal dependencies, and offers both C++ and Java interfaces for integration.

0

The size of TRIE is particularly sensitive to F1092 c11, so vocabulary filtering is quite effective at reducing model size.

The authors cluster NE instance pairs based on the words in the context using bag-of-words methods.

0

Automatic Paraphrase Discovery based on Context and Keywords between NE Pairs

Evaluation was done automatically using the BLEU score and manually on fluency and adequacy.

0

Often, two systems can not be distinguished with a confidence of over 95%, so there are ranked the same.

Human judges also pointed out difficulties with the evaluation of long sentences.

0

Since different judges judged different systems (recall that judges were excluded to judge system output from their own institution), we normalized the scores.

These clusters are computed using an SVD variant without relying on transitional structure.

0

In total there are O(K 2) parameters associated with the transition parameters.

It also incorporates the Good-Turing methodin estimating the likelihoods of previously unseen constructions, including morphological derivatives and personal names.

0

paper, and is missing 6 examples from the A set.

The PROBING data structure uses linear probing hash tables and is designed for speed.

0

IRSTLM 5.60.02 (Federico et al., 2008) is a sorted trie implementation designed for lower memory consumption.

Human judges also pointed out difficulties with the evaluation of long sentences.

0

The confidence intervals are computed by bootstrap resampling for BLEU, and by standard significance testing for the manual scores, as described earlier in the paper.

Das and Petrov, in this paper, approached inducing unsupervised part-of-speech taggers for languages that had no labeled training data, but had translated text in a resource-rich language.

0

Supervised learning approaches have advanced the state-of-the-art on a variety of tasks in natural language processing, resulting in highly accurate systems.

Using a treebank grammar, the data-driven lexicon, and a linguistically motivated unknown-tokens handling technique their model outperformed previous pipelined, integrated or factorized systems for Hebrew morphological and syntactic processing, yielding an error reduction of 12% over the best published results so far.

0

Tsarfaty (2006) used a morphological analyzer (Segal, 2000), a PoS tagger (Bar-Haim et al., 2005), and a general purpose parser (Schmid, 2000) in an integrated framework in which morphological and syntactic components interact to share information, leading to improved performance on the joint task.

From a DP-based solution to the traveling salesman problem, they present a novel technique to restrict the possible word reordering between source and target language in order to achieve an eÃcient search algorithm.

0

For the experiments, we use a simple preprocessing step.

They found replacing it with a ranked evaluation to be more suitable.

0

The easiest language pair according to BLEU (English-French: 28.33) received worse manual scores than the hardest (English-German: 14.01).

They extended previous work on discriminative weighting by using a finer granularity, focusing on the properties of instances rather than corpus components, and used simpler training procedure.

0

Finally, we make some improvements to baseline approaches.

they evaluate various specific aspects of the segmentation, as well as the overall segmentation performance.

0

In Table 5 we present results from small test corÂ pora for the productive affixes handled by the current version of the system; as with names, the segmentation of morphologically derived words is generally either right or wrong.

In order to handle the necessary word reordering as an optimization problem within the dynamic programming approach, they describe a solution to the traveling salesman problem (TSP) which is based on dynamic programming.

0

Two subjects are each given a calendar and they are asked to schedule a meeting.

This paper talks about Unsupervised Models for Named Entity Classification.

0

(6), with W+ > W_.

Using less training data than other systems, their NER can perform as well as other state-of-the-art NERs.

0

Note that we check for , the word preceding the consecutive sequence of initCaps tokens, since person prefixes like Mr., Dr., etc are not part of person names, whereas corporate suffixes like Corp., Inc., etc are part of corporate names.

Other kinds of productive word classes, such as company names, abbreviations,and place names can easily be handled given appropriate models.

0

Word type N % Dic tion ary entr ies 2 , 5 4 3 9 7 . 4 7 Mor pho logi call y deri ved wor ds 3 0 . 1 1 Fore ign tran slite rati ons 9 0 . 3 4 Per son al na mes 5 4 2 . 0 7 cases.

The PROBING data structure uses linear probing hash tables and is designed for speed.

0

We implement two data structures: PROBING, designed for speed, and TRIE, optimized for memory.

they evaluate various specific aspects of the segmentation, as well as the overall segmentation performance.

0

For derived words that occur in our corpus we can estimate these costs as we would the costs for an underived dictionary entry.

The use of global features has shown excellent result in the performance on MUC-6 and MUC-7 test data.

0

The features we used can be divided into 2 classes: local and global.

The authors believe automatic paraphrase discovery is an important component for building a fully automatic information extraction system.

0

Find keywords for each NE pair When we look at the contexts for each domain, we noticed that there is one or a few important words which indicate the relation between the NEs (for example, the word âunitâ for the phrase âa unit ofâ).

In this paper, Ben and Riloff present a coreference resolver called BABAR that focuses on the use of contextual-role knowledge for coreference resolution.

0

For example, if {N P1, N P2, N P3} are all coreferent, then each NP must be linked to one of the other two NPs.

The departure from the traditional token-based tagging approach allow them to explicitly capture type-level distributional properties of valid POS tag assignments as part of the model.

0

Once the lexicon has been drawn, the model proceeds similarly to the standard token-level HMM: Emission parameters Î¸ are generated conditioned on tag assignments T . We also draw transition parameters Ï.

The AdaBoost algorithm was developed for supervised learning.

0

2.

The corpus was annoted with different linguitic information.

0

Clearly this poses a number of research challenges, though, such as the applicability of tag sets across different languages.

This paper talks about KenLM: Faster and Smaller Language Model Queries.

0

This makes memory usage comparable to our PROBING model.

The manual evaluation of scoring translation on a graded scale from 1â5 seems to be very hard to perform.

0

This work was supported in part under the GALE program of the Defense Advanced Research Projects Agency, Contract No.

Finally, several coreference systems have successfully incorporated anaphoricity determination modules.

0

Table 1: Syntactic Seeding Heuristics BABARâs reliable case resolution heuristics produced a substantial set of anaphor/antecedent resolutions that will be the training data used to learn contextual role knowledge.

This paper offers a broad insight into of Arabic constituency parsing by analyzing the interplay of linguistic phenomena, annotation choices, and model design.

0

The ATB annotation guidelines specify that proper nouns should be specified with a flat NP (a).

This paper offers a broad insight into of Arabic constituency parsing by analyzing the interplay of linguistic phenomena, annotation choices, and model design.

0

Also surprising is the low test set OOV rate given the possibility of morphological variation in Arabic.

The PROBING data structure uses linear probing hash tables and is designed for speed.

0

Several methods exist to handle collisions; we use linear probing because it has less memory overhead when entries are small.

In order to create good-sized vectors for similarity calculation, they had to set a high frequency threshold.

0

This overview is illustrated in Figure 1.

they evaluate various specific aspects of the segmentation, as well as the overall segmentation performance.

0

Mandarin exhibits several such processes, including A-not-A question formation, ilÂ lustrated in (3a), and adverbial reduplication, illustrated in (3b): 3.

Finally, several coreference systems have successfully incorporated anaphoricity determination modules.

0

5 Related Work.

Bean and Riloff also used bootstrapping to extend their semantic compatibility model, proposed using caseframe network for anaphora resolution, information extraction patterns to identify contextual clues for determining compatibility between NPs.

0

Table 1 briefly describes the seven syntactic heuristics used by BABAR to resolve noun phrases.

The authors use linguistic and annotation insights to develop a manually annotated grammar and evaluate it and finally provide a realistic evaluation in which segmentation is performed in a pipeline jointly with parsing.

0

Instead, we extend the variation n-gram method of Dickinson (2005) to compare annotation error rates in the WSJ and ATB.