Boundary-based MWE segmentation with text partitioning
This work presents a fine-grained, text-chunking algorithm designed for the task of multiword expressions (MWEs) segmentation. As a lexical class, MWEs include a wide variety of idioms, whose automatic identification are a necessity for the handling of colloquial language. This algorithm’s core novelty is its use of non-word tokens, i.e., boundaries, in a bottom-up strategy. Leveraging boundaries refines token-level information, forging high-level performance from relatively basic data. The generality of this model’s feature space allows for its application across languages and domains. Experiments spanning 19 different languages exhibit a broadly-applicable, state-of-the-art model. Evaluation against recent shared-task data places text partitioning as the overall, best performing MWE segmentation algorithm, covering all MWE classes and multiple English domains (including user-generated text). This performance, coupled with a non-combinatorial, fast-running design, produces an ideal combination for implementations at scale, which are facilitated through the release of open-source software.
Jake Ryland Williams Drexel University 30 N. 33rd Street Philadelphia, PA 19104 firstname.lastname@example.org
Multiword expressions (MWEs) constitute a mixed class of complex lexical objects that often behave in syntactically unruly ways. A unifying property that ties this class together is the lexicalization of multiple words into a single unit. MWEs are generally difficult to understand through grammatical decomposition, casting them as types of minimal semantic units. There is variation in this non-compositionality property (Bannard et al., 2003), which in part may be attributed to differences in MWE types. These range from multiword named entities, such as Long Beach, California, to proverbs, such as it takes one to know one, to idiomatic verbal expressions, like cut it out (which often contain flexible gaps). For all of their strangeness they appear across natural languages (Jackendoff, 1997; Sag et al., 2002), though generally not for common meanings, and frequently with opaque etymologies that confound non-native speakers.
There are numerous applications in NLP for which a preliminary identification of MWEs holds great promise. This notably includes idiom-level machine translation (Carpuat and Diab, 2010); reduced polysemy in sense disambiguation (Finlayson and Kulkarni, 2011); keyphrase-refined information retrieval (Newman et al., 2012); and the integration of idiomatic and formulaic language in learning environments (Ellis et al., 2008). Parallel to these linguistically-focused applications is the possibility that MWE identification can positively affect machine learning applications in text analysis. Regardless of algorithm complexity, a common preliminary step in this area is tokenization. Having the “correct” segmentation of a text into words and MWEs results in a meaning-appropriate tokenization of minimal semantic units. Partial steps in this direction have been taken through recent work focusing on making the bag of phrases framework available as a simple improvement to the bag of words. However, that work (Handler et al., 2016) utilized only noun phrases, leaving the connection between MWEs and a comprehensive bag of phrases framework yet to be acknowledged. With the specific focus of MWEs on idiomaticity, a comprehensive bag of words and phrases framework would be possible, provided the MWE identification task is resolved.
1.2 Task description
Despite the variety that exist, studies often only focus on a few MWEs classes, or on only specific lengths (Tsvetkov and Wintner, 2011). In fact, named entity extraction may be thought of as satisfying the MWE identification task for just this one MWE class. The problem has a broader framing when all classes of MWEs are considered. Furthermore, since a mixed tokenization of words and phrases as minimal semantic units is a desired outcome, it is helpful to consider this task as a kind of fine-grained segmentation. Thus, this work refers to its task as MWE segmentation, and not identification or extraction. In other words, the specific goal here is to delimit texts into the smallest possible, independent units of meaning. Schneider et al. () were the first to treat this problem as such, when they created the first data set comprehensively annotated for MWEs. From this data set, an exemplar annotated record is:111 Note that color/indices redundantly indicate separate MWEs, with the colored box highlighting an MWE’s gap, and black, unnumbered text tokenized simply as words.
My wife had taken1 in1 for a routine oil3 change3.
whose segmentation is an example of the present focus of this work. Note that the present study focuses only on MWE tokens, does not aim to approach the task of MWE class identification, and does not attempt to disambiguate MWE meanings. For detailed descriptions of these other MWE-related tasks, Baldwin and Kim (2010) provide an extensive discussion.
1.3 Existing work
The identification of MWEs and collocations is an area of study that has seen notable focus in recent years (Seretan, 2008; Pecina, 2010; Newman et al., 2012; Ramisch, 2015; Schneider et al., 2014), and has a strong history of attention (both directly and through related work) in the literature (Becker, 1975; Church and Hanks, 1990; Sag et al., 2002). It has become commonplace for approaches to leverage well-studied machine learning algorithms such as structured perceptrons (Schneider et al., 2014) and conditonal random fields (Constant and Sigogne, 2011; Hosseini et al., 2016). The flexibility of these algorithms allow researchers to mix a variety of feature types, ranging from tokens to parts of speech to syntax trees. Juxtaposed to these relatively-complex models exist the simpler and more-heuristic (Cordeiro et al., 2015). Some rely singularly on MWE dictionaries, while others incorporate multiple measures or are rule-based, like those present in the suite available through mwetoolkit (Ramisch, 2015) or jMWE (Kulkarni and Finlayson, 2011).
MWEs have been the focus of considerable attention for languages other than English, too. Hungarian MWE corpora focusing on light verb constructions have been under development for some time (T. et al., 2011). In application to the French language, part-of-speech tagging has seen benefit (Constant and Sigogne, 2011) through awareness and relativity to MWEs. Recently, Savary et al. (2017) conducted a shared task for the identification of verbal MWEs with a data set spanning 18 languages (excluding English). While extending this area of work to a large variety of languages, this task saw notable multilingual algorithmic developments (Saied and Candito, 2017), but did not approach the identification of all MWE classes, comprehensively. On the other hand, a SemEval 2016 shared task (Schneider et al., 2016) covered English domains and all MWE classes, bearing the greatest similarity to the present work. In general, these shared tasks have all highlighted a need for the improvement of algorithms.
2.1 Text partitioning
Text partitioning is a physical model developed recently (Williams et al., 2015) for fine-grained text segmentation. It treats a text as a dichotomous squence, alternating between word () and non-word () tokens:
The key feature of text partitioning is its treatment of non-word, i.e., “boundary”, tokens. Acting like glue, these may take one of two distinct states, , identifying if a non-word token is bound () or broken (). A non-word token in the bound state binds words together. Thus, a text partitioning algorithm is a function that determines the states of non-word tokens.
In its original development, text partitioning was studied simplistically, with space as the only non-word token. In that work, a threshold probability, , was set. For each space, , in a text, a uniform random binding probability, , would be drawn. If , would be bound, and otherwise it would be broken. As a parameter, thus allowed for the tuning of a text into its collection of words (), clauses (), or, for any value, , a randomly-determined collection of -grams. While non-deterministic, this method was found to preserve word frequencies, (unlike the sliding-window method), and made possible the study of Zipf’s law for mixed distributions of words and -grams.
The present work utilizes the parameter to develop a supervised machine learning algorithm for MWE segmentation. A threshold probability, , is still set, and the supervised component is the determination of the binding probabilities () for a text’s non-word tokens. Provided a gold-standard, MWE-segmented text:
let denote the frequency at which a boundary is observed between and in the state . Provided this, a binding probability is defined as:
This basic, -gram text partitioning model makes the binding probabilities a function of boundaries and their immediately-surrounding words. In principle, this might be extended to a more-nuanced model, with binding probabilities refined by larger-gram information.
Some MWEs consist of non-contiguous spans of words. These varieties are often referred to as “gappy” expressions, an example of which is shown in Sec. 1.2. Text partitioning may easily be extended to handle gappy MWEs by instituting a unique boundary token, e.g.,
that indicates the presence of a gap. Since gappy MWEs are relatively sparse as compared to other MWEs, a single gap-boundary token is used for all gap sizes. This is designed for a flexible handling of variable gap sizes, given the relatively small amount of gold-standard data that is presently available. However, this may in principle be refined to particular gap-sized specifications, possibly ideal for higher precision in the presence of larger quantities of gold-standard data.
A number of MWE types, such as named entities, are entirely open classes. Often occurring only once, or as entirely emergent objects, these pose a significant challenge for MWE segmentation, along with the general sparsity and size of the current gold-standards. For their inclusion in the gold-standard datasets and the general quality of automated taggers, part-of-speech (POS) information may generally be leveraged to increase recall. These data are utilized in a parallel text partitioning algorithm, swapping tokens for tags.222 Note that this requires the inclusion of a special POS tag, e.g., “SP”, for the space character. Via two independent thresholds, and , the combined algorithm merges candidate MWEs.
2.2 The longest first defined
In the presented form, text partitioning only focuses on information immediately local to boundaries (surrounding word pairs). This has positive effects for recall, but can result in lower precision, since there is no guarantee that a sequence of bound tokens is an MWE. For example, if presented with the text:
“I go for take out there, frequently.”
the segment take out there might be bound, since take out and out there are both known MWE forms, potentially observed in training. To balance this, a directional, lookup-based algorithm is proposed. Referred to as the longest first defined (LFD) algorithm (see Alg. 1), this algorithm prunes candidates by clipping off the longest known (MWE) references along the reading direction of a language. This requires knowledge of MWE lexica, which may be derived from both gold-standard data and external sources (see Sec. 3). Continuing with the example, if the text partitioning algorithm outputs the candidate, take out there, it would next be passed to the LFD. The LFD would find take out there unreferenced, and check the next-shortest (2-word) segments, from left to right. The LFD would immediately find take out referenced, output it, and continue on the remainder, there. With only one term remaining, the word there would then be trivially output and the algorithm terminated. While this algorithm will likely fail when confronted with pathological expressions, like those in “garden path” sentences, e.g., “The prime number few.”, directionality is a powerful heuristic in many languages that may be leveraged for increased precision.
3.1 Gold standard data
Treating MWE segmentation as a supervised machine learning task, this work relies on several recently-constructed MWE-annotated data sets. This includes the business reviews contained in the Supersense-Tagged Repository of English with a Unified Semantics for Lexical Expressions, annotated by Schneider et al. Schneider et al. (); Schneider and Smith (2015). These data were harmonized and merged with the Ritter and Lowlands data set of supersense-annotated tweets (Johannsen et al., 2014) for the SemEval 2016 shared task (#10) on Detecting Minimal Semantic Units and their Meanings (DIMSUM), conducted by Schneider et al. (2016). The DIMSUM data set additionally possesses token lemmas and gold-standard part of speech (POS) tags for the 17 universal POS categories. In addition to the shared task training data of business reviews and tweets, the DIMSUM shared task resulted in the creation of three domains of testing data, which spanned business reviews, tweets, and TED talk transcripts. All DIMSUM data are comprehensive in being annotated for all MWE classes.
To evaluate against a diversity of languages this work also utilizes data produced by the multinational, European Cooperation in Science and Technology’s action group: PARSing and Multiword Expressions within a European multilingual network (PARSEME) (Savary et al., 2015). In 2017, the PARSEME group conducted a shared task with data spanning 18 languages333 While the shared task was originally planned to cover 21 languages, corpus release was only achieved for Bulgarian (BG), Czech (CS), German (DE), Greek (EL), Spanish (ES), Farsi (FA), French (FR), Hebrew (HE), Hungarian (HU), Italian (IT), Lithuanian (LT), Maltese (MT), Polish (PL), Brazilian Portuguese (PT), Romanian (RO), Slovene (SL), Swedish (SV), and Turkish (TR). No sufficiently available native annotators were found for English (EN), Yiddish (YI), and Croatian (HR). High-level data (including POS tags) were provided for all of the 18 languages, except BG, HE, and LT. (Savary et al., 2017), focusing on several classes of verbal MWEs. So, while the PARSEME data are not annotated for all MWEs classes, they do provide an assessment against multiple languages. However, the resources gathered for the 18 languages exhibit a large degree of variation in overall size and numbers of MWEs annotated, leading to observable differences in identifiability.
The gold standard data sets were produced with variations in annotation formats. The DIMSUM data set utilizes a variant of the beginning inside outside (BIO) scheme (Ramshaw and Marcus, 1995) used for named entity extraction. Additionally, their annotations indicate which tokens are linked to which, as opposed to the PARSEME data set, which simply identifies tokens to indexed MWEs. Note that this has implications to task evaluation: the PARSEME evaluations can only assess tokens’ presence inside of specific MWEs, while the DIMSUM evaluations can focus on specific token-token attachments/separations. Evaluations against the DIMSUM datasets are therefore more informative of segmentation, than identification. Additionally, the DIMSUM data sets use lowercase BIO tags to indicate the presence of tokens inside of the gaps of others. However, the DIMSUM data sets provide no information on the locations of spaces in sentences, unlike the PARSEME data sets, which do. Since the present work relies on knowledge of spaces to identify token-token boundaries for segmentation, the DIMSUM data sets had to first be pre-processed to infer the locations of spaces. This is done in such a way as to preserve comparability with the work others, (discussed in Sec. 4.1).
3.2 Support data
The gold-standard data sets (DIMSUM, and PARSEME) exhibit variations in size, domain, language, and in the classes of annotated MWEs. Ideally, each of these data sets would cover all MWE classes. Since the English data sets do, and many are open classes (e.g., the named entity class readily accepts new members), gold standards cannot be expected to cover all MWE forms. So, to produce segmentations that identify rare MWEs, like those that occur once in the gold standard data, this work relies on support data. Note that because the PARSEME data set covers a restricted set of MWE types (verbal MWEs, only), it would likely not help to incorporate external sources. Thus, the support data described below are only used for the English language experiments, i.e., the DIMSUM data sets.
Since this work approaches the problem as a segmentation task, information is needed on MWE edge-boundaries. Thus, support data must present MWEs in their written contexts, and not just as entries in a lexicon. Example usages of dictionary entries provide this detail, and are leveraged from Wiktionary (data accessed 1/11/16) and Wordnet (Miller, 1995). These exemplified dictionary entries help to fill gold standard data gaps, but still lack many noun compounds and named entities. Outside of dictionaries, MWEs such as these may be found in encyclopedias. Thus, the Wikipedia hyperlinks present in all Wikipedia (data accessed 5/1/16) articles are utilized. Specifically, the exact hyperlink targets are used (not the displayed text), and without using any term extraction measures for filtering, as opposed to the data produced by Hartmann et al. (2012). This results in data that are noisy, with many entities that may not actually be classifiable as MWEs. However, their availability and broad coverage offset these negative properties, which is exhibited by this work’s evaluation.
None of the gold standard data sets explicitly identify the locations of spaces in their annotations. This is a challenge for the present work, since it focuses on word-word boundaries (of which space is the most common) to identify the separations between segments. This turns out to not be an issue with the PARSEME data sets, which indicate when a given token is not followed by a space. However for the DIMSUM data sets, the locations of spaces had to be inferred. To resolve this issue, a set of heuristic rules are adopted with a default assumption of space on both sides of a tokens. Exceptions to this default include, group openings (e.g., brackets and parentheses) and odd-indexed quotes (double, single, etc.), for which space is only assumed at left; and punctuation tokens (e.g., commas and periods), group closures (e.g., brackets and parentheses), and even-indexed quotes (double, single, etc.), for which space is only assumed at right. While these heuristics will certainly not correctly identify all instances of space, they make the data sets more faithful to their original texts. Furthermore, since the annotations and evaluation procedures only focus on links between non-space tokens, the data may be re-indexed during pre-processing so as to allow for any resulting evaluation to be comparable to those of the data set authors’ and shared task participants’. Thus, the omission of space characters and their inference in this work only negatively impacts text partitioning’s evaluation. In other words, if this work were applied to annotated data that properly represents space, higher performance might be exhibited.
It is reasonably straightforward to measure precision, recall, and for exact matches of MWEs. However, this strategy is unreasonably coarse, failing to represent partial credit when algorithms get only portions of MWEs correct. Thus, the developers of the different gold standard data sets have established other evaluation metrics that are more flexible. Utilizing these partial credit MWE evaluation metrics provides refined detail into the performance of algorithms. However, these are not the same across the gold standard data sets. So, to maintain comparability of the present results, this work uses the specific strategies associated to each shared task.
In application to the PARSEME data sets, precision, recall, and describe tokens’ presence in MWEs. Alternatively, DIMSUM-style metrics measure link/boundary-based evaluations. Specifically, this strategy checks if the links between tokens are correct. Note that this latter (DIMSUM) evaluation is better aligned to the formulation of text partitioning, but leaves the number evaluation points at one fewer per MWE than the PARSEME scheme. Thus, PARSEME evaluations favor longer MWEs more heavily.
|EN||N||0.5||0.71||0.5396||0.5507||0.5451||3/5||0.1348 – 0.5724|
|Tweets||N||0.5||0.71||0.5897||0.5226||0.55542||3/5||0.1550 – 0.6109|
|Reviews||N||0.5||0.71||0.5721||0.5584||0.5626||1/5||0.0868 – 0.5408|
|TED||N||0.5||0.71||0.3984||0.6108||0.4823||3/5||0.2011 – 0.5714|
|BG||N||0.79||N/A||0.7071||0.5141||0.5954||2/3||0.5916 – 0.6615|
|CS||N||0.73||0.0||0.7849||0.6655||0.7203||3/5||0.2352 – 73.65|
|DE||N||0.82||0.0||0.5582||0.2788||0.3719||4/6||0.283 – 0.4545|
|EL||N||0.78||0.0||0.3931||0.3815||0.3872||5/6||0.3871 – 0.4688|
|ES||N||0.64||0.71||0.7473||0.4098||0.5293||2/6||0.3093 – 0.5839|
|FA||N||0.57||0.68||0.7040||0.8313||0.7624||3/3||0.8536 – 0.9020|
|FR||N||0.73||0.0||0.6589||0.3836||0.4849||4/7||0.1 – 0.6152|
|HE||N||0.78||N/A||0.5969||0.2107||0.3115||2/3||0.0 – 0.313|
|HU||N||0.97||0.66||0.7221||0.6612||0.6903||2/6||0.6226 – 0.7081|
|IT||N||0.85||0.0||0.5497||0.3174||0.4024||2/5||0.1824 – 0.4357|
|LT||N||0.79||N/A||0.6567||0.1803||0.2830||1/3||0.0 – 0.2533|
|MT||N||0.86||0.0||0.1591||0.1538||0.1564||2/5||0.0 – 0.1629|
|PL||N||0.66||0.0||0.8962||0.5966||0.7164||2/5||0.0 – 0.7274|
|PT||N||0.79||0.0||0.7518||0.4921||0.5948||4/5||0.3079 – 0.7094|
|RO||N||0.71||0.0||0.8350||0.7850||0.8092||3/5||0.7799 – 0.8358|
|SL||N||0.7||0.0||0.6606||0.4504||0.5356||1/5||0.3320 – 0.4655|
|SV||N||1.0||0.95||0.0949||0.7771||0.1691||5/5||0.2669 – 0.3149|
|TR||N||0.87||0.0||0.3852||0.3706||0.3778||5/5||0.4550 – 0.5528|
4.3 Experimental design
The basic text partitioning model relies on the single threshold parameter, , and integration of POS tags relies on a second. So, optimization ultimately entails the determination of parameters for both tokens, , and and POS tags . To balance both precision and recall, these parameters are determined through optimization of the measure. In the absence of the LFD, -optimal pairs, , are first determined via a full parameter scan over
For a given threshold pair, LFD-enhancement can then only increase precision, while decreasing recall. So, subsequent optimization with the LFD is accomplished through scanning values of and in the parameter space no less than those previously determined for basic, non-LFD model.
The different experiments were conducted in accordance with the protocols established by the designers of data sets and shared tasks, and in all cases, an eight-fold cross-validation was conducted for optimization. Exact comparability was achieved for the DIMSUM and PARSEME experiments as a result of the precise configurations of training and testing data from the shared tasks. Moreover, since an evaluation script was provided for each, metrics reported for DIMSUM and PARSEME experiments are in complete accord with the results of the shared tasks. For the DIMSUM experiments, results should be compared to the open track (external data was utilized), and for the PARSEME experiments, results should be compared to the closed track (no external data was utilized).
Evaluations spanning the variety of languages (19, in total) showed high levels of performance, especially in application to English, where there was a diversity of domains (business reviews, Tweets, and TED talk transcripts), along with comprehensive MWE annotations. Moreover, these results were generally observed for text partitioning both with, and without the LFD. As expected, application of the LFD generally led to increased precision. While integration of POS tags was found to generally improve MWE segmentation in all English experiments, this was frequently not the case in applications to other languages. However, this observation should be taken with consideration for the restriction to the fewer MWE classes (verbal MWEs, only) annotated in the PARSEME (non-English) shared task languages, and additionally the fact that no external data were used. Detailed results for all DIMSUM and PARSEME experiments are recorded in Tab. 1.
For the DIMSUM experiments, final parameterizations were determined as for text partitioning, alone, and for the LFD-enhanced model. Comparing the base and LFD-enhanced models, higher overall performance was always achieved with the LFD (increasing by as many as 12 points). Including text partitioning in the shared-task rankings (for a total of 5 models) placed the LFD-enhanced model first at all domains but Twitter, for which third was reached (though within 3 -points of first). However, combining all three domains into a single experiment placed the LFD-enhanced text partitioning algorithm as first, making it the best-performing algorithm, overall. In application to the user-reviews domain, text partitioning maintained first-place status, even without the LFD enhancement. For all other domains the base model ranked third.
For the PARSEME experiments, final parameterizations varied widely. This is not surprising, considering the significant variation in data set annotations and domains across the 18 languages. Additionally, POS tags were found to be of less-consistent value to the text partitioning algorithm, particularly when the LFD was not applied. Indeed, cross-validation of the base model resulted in as optimal for 11 out of the 15 languages where POS tags were made available. However, cross-validation of the LFD-enhanced algorithm resulted in only 6 parameterizations having as optimal. First place status was achieved for three out of the 18 languages (LT, PL, and SL), and for all languages aside from SV and TR, mid-to-high ranking values were achieved.444 Note that anomalous MWEs were observed in the DE HU data sets, where large portions of the annotated MWEs consisted of only a single token. While the PARSEME annotation scheme includes multiword components that span a single token, e.g., “don’t” in don’t talk the talk, those observed in DE and HU were found outside of the annotation format. This included of all MWEs annotated in the DE test records and of all in the HU test records. Since text partitioning identifies segment boundaries, it cannot handle these anomalous MWEs, unlike the models entered into the PARSEME shared task. So to accommodate these and maintain comparability, a separate algorithm was employed. This simply placed lone MWE tags on tokens that were observed as anomalous or more of the time in training. In contrast to the DIMSUM data sets, application of the LFD improved scores in only roughly half of the experiments.
Evaluation against the comprehensively-annotated English data sets has shown text partitioning to be the current highest overall ranking MWE segmentation algorithm. This result is upheld for two out of the three available test domains (business reviews and TED talk transcripts), with a close third place achieved against data from Twitter. This exhibits the algorithms general applicability across domains, and especially in the context of noisy text. Combined with the algorithm’s fast-running and non-combinatorial nature, this makes text partitioning ideal for large-scale applications to the identification of colloquial language, often found on social media. For these purposes, the presented algorithms have been made available as open-source tools as the Python “Partitioner” module, which may be accessed through Github555 https://github.com/jakerylandwilliams/partitioner and the Python Package Index666 https://pypi.python.org/pypi/partitioner for general use.
Unfortunately, the PARSEME experiments did not provide an evaluation against all types of MWEs. However, they did exhibit the general applicability of text partitioning across languages. So, while the PARSEME data are not sufficient for comprehensive MWE segmentation, trained models have also been made available for the 18 non-english languages through the Python Partitioner module. Across the 18 PARSEME shared-task languages text partitioning’s values were found to rank as mid to high, with the notable exception of SV. However, the SV data is peculiar in being quite small, with its training set smaller than its testing set. However, models entered into the PARSEME shared task achieved roughly twice the score for SV, indicating the possibility that text partitioning requires some critical mass of training data in order to achieve high levels of performance. Thus, for general increases in performance and for extension to comprehensive MWE segmentations, future directions of this work will likely do well to seek the collection of larger and more-comprehensive data sets.
As defined, text partitioning is subtly different from a 2-gram model: it focuses on non-word boundary tokens, as opposed to just word-word pairs. Because this algorithm relies on knowledge of boundary token states, it cannot be trained well on MWE lexica, alone. Fort this model to achieve high precision, boundaries commonly occurring as broken must be observed as such, even if they are necessary components of known MWEs. Thus, the use of boundary-adjacent words for prediction is a limitation of the present model. This may possibly be overcome through use of more distant words and boundaries. However, since gold-standard data are still relatively small, they will likely require significant expansion before such models may be effectively implemented. Thus, future directions with more nuanced text partitioning models highlight the importance of generating more gold standard data, too.
- Baldwin and Kim (2010) Timothy Baldwin and Su Nam Kim. 2010. Muiltiword expressions. In Handbook of Natural Language Processing. CRC Press, Boca Raton, USA, pages 267–292.
- Bannard et al. (2003) Colin Bannard, Timothy Baldwin, and Alex Lascarides. 2003. A statistical approach to the semantics of verb-particles. In Proceedings of the ACL 2003 Workshop on Multiword Expressions: Analysis, Acquisition and Treatment - Volume 18. pages 65–72.
- Becker (1975) Joseph D. Becker. 1975. The phrasal lexicon. In Proceedings of the 1975 Workshop on Theoretical Issues in Natural Language Processing. TINLAP ’75, pages 60–63.
- Carpuat and Diab (2010) Marine Carpuat and Mona Diab. 2010. Task-based evaluation of multiword expressions: A pilot study in statistical machine translation. In Human Language Technologies: The 2010 Annual Conference of the North American Chapter of the Association for Computational Linguistics. HLT ’10, pages 242–245.
- Church and Hanks (1990) Kenneth W. Church and Patrick Hanks. 1990. Word association norms, mutual information, and lexicography. Computational Linguistics 16(1):22–29. Http://dl.acm.org/citation.cfm?id=89086.89095.
- Constant and Sigogne (2011) Matthieu Constant and Anthony Sigogne. 2011. Mwu-aware part-of-speech tagging with a crf model and lexical resources. In Proceedings of the Workshop on Multiword Expressions: From Parsing and Generation to the Real World. MWE ’11, pages 49–56.
- Cordeiro et al. (2015) Silvio Ricardo Cordeiro, Carlos Ramisch, and Aline Villavicencio. 2015. Token-based mwe identification strategies in the mwetoolkit. In Proceedings of the PARSEME 4th general meeting.
- Ellis et al. (2008) Nick C. Ellis, Rita Simpson-Vlach, and Carson Maynard. 2008. Formulaic Language in Native and Second Language Speakers: Psycholinguistics, Corpus Linguistics, and TESOL. TESOL Quarterly: A Journal for Teachers of English to Speakers of Other Languages and of Standard English as a Second Dialect 42(3):375–396.
- Finlayson and Kulkarni (2011) Mark Alan Finlayson and Nidhi Kulkarni. 2011. Detecting multi-word expressions improves word sense disambiguation. In Proceedings of the Workshop on Multiword Expressions: From Parsing and Generation to the Real World. MWE ’11, pages 20–24.
- Handler et al. (2016) Abram Handler, Matthew J. Denny, Hanna Wallach, and Brendan OâConnor. 2016. Bag of what? simple noun phrase extraction for text analysis. In Proceedings of the Workshop on Natural Language Processing and Computational Social Science. pages 114–124.
- Hartmann et al. (2012) Silvana Hartmann, György Szarvas, and Iryna Gurevych. 2012. Mining multiword terms from wikipedia pages 226–258.
- Hosseini et al. (2016) Mohammad Javad Hosseini, Noah A. Smith, and Su-In Lee. 2016. UW-CSE at SemEval-2016 Task 10: Detecting multiword expressions and supersenses using double-chained conditional random fields. In Proceedings of SemEval.
- Jackendoff (1997) Ray S. Jackendoff. 1997. The Architecture of the Language Faculty. MIT Press.
- Johannsen et al. (2014) Anders Johannsen, Dirk Hovy, HÃ©ctor Martinez, Barbara Plank, and Anders SÃ¸gaard. 2014. More or less supervised super-sense tagging of twitter. In The 3rd Joint Conference on Lexical and Computational Semantics.
- Kulkarni and Finlayson (2011) Nidhi Kulkarni and Mark Alan Finlayson. 2011. jmwe: A java toolkit for detecting multi-word expressions. In Proceedings of the Workshop on Multiword Expressions: From Parsing and Generation to the Real World. MWE ’11, pages 122–124.
- Miller (1995) George A. Miller. 1995. Wordnet: A lexical database for english. Commun. ACM 38(11):39–41.
- Newman et al. (2012) David Newman, Nagendra Koilada, Jey Han Lau, and Timothy Baldwin. 2012. Bayesian text segmentation for index term identification and keyphrase extraction. In Proceedings of the 9th Workshop on Multiword Expressions. pages 139–144.
- Pecina (2010) Pavel Pecina. 2010. Lexical association measures and collocation extraction. Language Resources and Evaluation 44(1-2):137–158. Http://dx.doi.org/10.1007/s10579-009-9101-4. https://doi.org/10.1007/s10579-009-9101-4.
- Ramisch (2015) Carlos Ramisch. 2015. Multiword Expressions Acquisition: A Generic and Open Framework. Springer Publishing Company, Incorporated. https://doi.org/10.1007/978-3-319-09207-2.
- Ramshaw and Marcus (1995) Lance A. Ramshaw and Mitchell P. Marcus. 1995. Text chunking using transformation-based learning. In Proceedings of the Third ACL Workshop on Very Large Corpora. pages 82–94.
- Sag et al. (2002) Ivan A. Sag, Timothy Baldwin, Francis Bond, Ann Copestake, and Dan Flickinger. 2002. Multiword Expressions: A Pain in the Neck for NLP. Springer, Berlin, Heidelberg. https://doi.org/10.1007/3-540-45715-1_1.
- Saied and Candito (2017) Hazem Al Saied and Marie Candito. 2017. The atilf-llf system for parseme shared task: a transition-based verbal multiword expression tagger. In Proceedings of the 13th Workshop on Multiword Expressions. pages 127–132.
- Savary et al. (2017) Agata Savary, Carlos Ramisch, Silvio Cordeiro, Federico Sangati, Veronika Vincze, Behrang QasemiZadeh, Marie Candito, Fabienne Cap, Voula Giouli, Ivelina Stoyanova, and Antoine Doucet. 2017. The parseme shared task on automatic identification of verbal multiword expression. In Proceedings of the 13th Workshop on Multiword Expressions. pages 31–47.
- Savary et al. (2015) Agata Savary, Manfred Sailer, Yannick Parmentier, Michael Rosner, Victoria Rosén, Adam Przepiórkowski, Cvetana Krstev, Veronika Vincze, Beata Wójtowicz, Gyri Smørdal Losnegaard, Carla Parra Escartín, Jakub Waszczuk, Matthieu Constant, Petya Osenova, and Federico Sangati. 2015. PARSEME – PARSing and Multiword Expressions within a European multilingual network. In 7th Language & Technology Conference: Human Language Technologies as a Challenge for Computer Science and Linguistics (LTC 2015).
- Schneider et al. (2014) Nathan Schneider, Emily Danchik, Chris Dyer, and Noah A. Smith. 2014. Discriminative lexical semantic segmentation with gaps: Running the mwe gamut. Transactions of the Association for Computational Linguistics 2:193–206.
- Schneider et al. (2016) Nathan Schneider, Dirk Hovy, Anders Johannsen, and Marine Carpuat. 2016. SemEval-2016 Task 10: Detecting Minimal Semantic Units and their Meanings (DiMSUM). In Proceedings of SemEval.
- (27) Nathan Schneider, Spencer Onuffer, Nora Kazour, Emily Danchik, Michael T. Mordowanec, Henrietta Conrad, and Noah A. Smith. ????
- Schneider and Smith (2015) Nathan Schneider and Noah A. Smith. 2015. A corpus and model integrating multiword expressions and supersenses. In Proceedings of the 2015 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies. The Association for Computational Linguistics, pages 1537–1547.
- Seretan (2008) V. Seretan. 2008. Collocation Extraction Based on Syntactic Parsing. Ph.D. thesis, University of Geneva.
- T. et al. (2011) István Nagy T., Gábor Berend, György Móra, and Veronika Vincze. 2011. Domain-dependent detection of light verb constructions. In RANLP Student Research Workshop. pages 1–8.
- Tsvetkov and Wintner (2011) Yulia Tsvetkov and Shuly Wintner. 2011. Identification of multi-word expressions by combining multiple linguistic information sources. In Proceedings of the Conference on Empirical Methods in Natural Language Processing. EMNLP ’11, pages 836–845.
- Williams et al. (2015) Jake Ryland Williams, Paul R. Lessard, Suma Desu, Eric M. Clark, James P. Bagrow, Chris M. Danforth, and Peter Sheridan Dodds. 2015. Zipf’s law holds for phrases, not words. Scientific Reports 5.