Multi-word Terms Research Articles

Supporting systematic reviews using LDA-based document representations.

BackgroundIdentifying relevant studies for inclusion in a systematic review (i.e. screening) is a complex, laborious and expensive task. Recently, a number of studies has shown that the use of machine learning and text mining methods to automatically identify relevant studies has the potential to drastically decrease the workload involved in the screening phase. The vast majority of these machine learning methods exploit the same underlying principle, i.e. a study is modelled as a bag-of-words (BOW).MethodsWe explore the use of topic modelling methods to derive a more informative representation of studies. We apply Latent Dirichlet allocation (LDA), an unsupervised topic modelling approach, to automatically identify topics in a collection of studies. We then represent each study as a distribution of LDA topics. Additionally, we enrich topics derived using LDA with multi-word terms identified by using an automatic term recognition (ATR) tool. For evaluation purposes, we carry out automatic identification of relevant studies using support vector machine (SVM)-based classifiers that employ both our novel topic-based representation and the BOW representation.ResultsOur results show that the SVM classifier is able to identify a greater number of relevant studies when using the LDA representation than the BOW representation. These observations hold for two systematic reviews of the clinical domain and three reviews of the social science domain.ConclusionsA topic-based feature representation of documents outperforms the BOW representation when applied to the task of automatic citation screening. The proposed term-enriched topics are more informative and less ambiguous to systematic reviewers.Electronic supplementary materialThe online version of this article (doi:10.1186/s13643-015-0117-0) contains supplementary material, which is available to authorized users.

Biomedical term extraction: overview and a new methodology

Terminology extraction is an essential task in domain knowledge acquisition, as well as for information retrieval. It is also a mandatory first step aimed at building/enriching terminologies and ontologies. As often proposed in the literature, existing terminology extraction methods feature linguistic and statistical aspects and solve some problems related (but not completely) to term extraction, e.g. noise, silence, low frequency, large-corpora, complexity of the multi-word term extraction process. In contrast, we propose a cutting edge methodology to extract and to rank biomedical terms, covering all the mentioned problems. This methodology offers several measures based on linguistic, statistical, graphic and web aspects. These measures extract and rank candidate terms with excellent precision: we demonstrate that they outperform previously reported precision results for automatic term extraction, and work with different languages (English, French, and Spanish). We also demonstrate how the use of graphs and the web to assess the significance of a term candidate, enables us to outperform precision results. We evaluated our methodology on the biomedical GENIA and LabTestsOnline corpora and compared it with previously reported measures.

Многокомпонентные термины в функции ключевых слов в информационной базе славянского языкознания iSybislaw

Multi-word terms as keywords in the iSybislaw databaseThis article is devoted to the problem of the criteria on which multi­-word terms should be selected and verified for the bibliographic database of Slavic linguistics publications iSybislaw. The text is based on Russian and Polish linguistic material. Such criteria as actuality, frequency, transparency of form are applied separately in each language and different sets and variability of terms in the each case are established. The authors examine also the reasons of formal and semantic asymmetry between Polish and Russian terminological units in the database.

A Hybrid Multi-Word Terms Extraction System Applied to Topic Detection

Mutli-word Terms extraction plays an important role in many Natural Language Processing (NLP) tasks. Despite their major importance, few works were dedicated to Arabic multi-word terms extraction. This paper proposes an automatic Arabic multi-word terms (MWTs) extraction system based on two major filtering steps: linguistics filter using a part-of-speech tagger along with morphological patterns and statistical filter based on probabilistic methods, namely: Log-Likelihood Ratio (LLR) and C-value. We evaluate the performances of the realized systems on Wattan; an Arabic oriented topic newspaper corpus. Our system manages to achieve 90.23% in term of multi-word extraction precision. We also study the use of MWTs as features in Arabic Topic Detection. The conducted experiments show good results.

A Novel Method for Arabic Multi-Word Term Extraction

Arabic Multiword Terms (AMWTs) are relevant strings of words in text documents. Once they are automatically extracted, they can be used to increase the performance of any Arabic Text Mining applications such as Categorization, Clustering, Information Retrieval System, Machine Translation, and Summarization, etc. Mainly the proposed methods for AMWTs extraction can be categorized in three approaches: Linguistic-based, Statistic-based, and hybrid-based approach. These methods present some drawbacks that limit their use. In fact they can only deal with bi-grams terms and their yield not good accuracies. In this paper, to overcome these drawbacks, we propose a new and efficient method for AMWTs Extraction based on a hybrid approach. This latter is composed by two main filtering steps: the Linguistic filter and the Statistical one. The Linguistic Filter uses our proposed Part Of Speech (POS) Tagger and the Sequence identifier as patterns in order to extract candidate AMWTs. While the Statistical filter incorporate the contextual information, and a new proposed association measure based on Termhood and Unithood Estimation named NTC-Value. To evaluate and illustrate the efficiency of our proposed method for AMWTs extraction, a comparative study has been conducted based on Kalimat Corpus and using nine experiment schemes: In the linguistic filter, we used three POS Taggers such as Taani’s method based Rule-approach, HMM method based Statistical-approach, and our recently proposed Tagger based Hybrid –approach. While in the Statistical filter, we used three statistical measures such as C-Value, NC-Value, and our proposed NTC-Value. The obtained results demonstrate the efficiency of our proposed method for AMWTs extraction: it outperforms the other ones and can deal correctly with the tri-grams terms.

Evaluation of five single-word term recognition methods on a legal English corpus

Specialised texts are characterised by, amongst other features, the presence of terminology which conveys domain-specific concepts that are essential for the specialist who is interested in analysing such texts. Automatic Term Recognition methods (ATR) are employed to identify those terms automatically, which is especially helpful in view of the large size of corpora nowadays. However, they tend to concentrate on the identification of Multi-Word Terms (MWTs) neglecting Single-Word Terms (SWTs) to a certain extent. This might be related to the greater number of the former found in fields such as biomedicine. However, so far as legal English is concerned, testing has shown that SWTs represent 65.22 percent of the items in the specialised glossary employed for the evaluation of the ATR methods examined herein. This paper presents the evaluation of five SWT recognition methods, namely, those of Chung (2003) , Drouin (2003) , Kit and Liu (2008) , Keywords (2008), and TF-IDF (term frequency-inverse document frequency). These were tested on the United Kingdom Supreme Court Corpus (UKSCC), a legal corpus of 2.6 million words which was compiled for this purpose. The results indicate that Drouin's TermoStat software is the best performing method, achieving 73.45 percent precision on the top 2,000 candidate terms.

An intelligent content discovery technique for health portal content management.

BackgroundContinuous content management of health information portals is a feature vital for its sustainability and widespread acceptance. Knowledge and experience of a domain expert is essential for content management in the health domain. The rate of generation of online health resources is exponential and thereby manual examination for relevance to a specific topic and audience is a formidable challenge for domain experts. Intelligent content discovery for effective content management is a less researched topic. An existing expert-endorsed content repository can provide the necessary leverage to automatically identify relevant resources and evaluate qualitative metrics.ObjectiveThis paper reports on the design research towards an intelligent technique for automated content discovery and ranking for health information portals. The proposed technique aims to improve efficiency of the current mostly manual process of portal content management by utilising an existing expert-endorsed content repository as a supporting base and a benchmark to evaluate the suitability of new contentMethodsA model for content management was established based on a field study of potential users. The proposed technique is integral to this content management model and executes in several phases (ie, query construction, content search, text analytics and fuzzy multi-criteria ranking). The construction of multi-dimensional search queries with input from Wordnet, the use of multi-word and single-word terms as representative semantics for text analytics and the use of fuzzy multi-criteria ranking for subjective evaluation of quality metrics are original contributions reported in this paper.ResultsThe feasibility of the proposed technique was examined with experiments conducted on an actual health information portal, the BCKOnline portal. Both intermediary and final results generated by the technique are presented in the paper and these help to establish benefits of the technique and its contribution towards effective content management.ConclusionsThe prevalence of large numbers of online health resources is a key obstacle for domain experts involved in content management of health information portals and websites. The proposed technique has proven successful at search and identification of resources and the measurement of their relevance. It can be used to support the domain expert in content management and thereby ensure the health portal is up-to-date and current.

Terminology extraction from medical texts in Polish.

BackgroundHospital documents contain free text describing the most important facts relating to patients and their illnesses. These documents are written in specific language containing medical terminology related to hospital treatment. Their automatic processing can help in verifying the consistency of hospital documentation and obtaining statistical data. To perform this task we need information on the phrases we are looking for. At the moment, clinical Polish resources are sparse. The existing terminologies, such as Polish Medical Subject Headings (MeSH), do not provide sufficient coverage for clinical tasks. It would be helpful therefore if it were possible to automatically prepare, on the basis of a data sample, an initial set of terms which, after manual verification, could be used for the purpose of information extraction.ResultsUsing a combination of linguistic and statistical methods for processing over 1200 children hospital discharge records, we obtained a list of single and multiword terms used in hospital discharge documents written in Polish. The phrases are ordered according to their presumed importance in domain texts measured by the frequency of use of a phrase and the variety of its contexts. The evaluation showed that the automatically identified phrases cover about 84% of terms in domain texts. At the top of the ranked list, only 4% out of 400 terms were incorrect while out of the final 200, 20% of expressions were either not domain related or syntactically incorrect. We also observed that 70% of the obtained terms are not included in the Polish MeSH.ConclusionsAutomatic terminology extraction can give results which are of a quality high enough to be taken as a starting point for building domain related terminological dictionaries or ontologies. This approach can be useful for preparing terminological resources for very specific subdomains for which no relevant terminologies already exist. The evaluation performed showed that none of the tested ranking procedures were able to filter out all improperly constructed noun phrases from the top of the list. Careful choice of noun phrases is crucial to the usefulness of the created terminological resource in applications such as lexicon construction or acquisition of semantic relations from texts.

A Corpus-based Machine Translation Method of Term Extraction in LSP Texts

To tackle the problems of term extraction in language specific field, this paper proposes a method of coordinating use of corpus and machine translation system in extracting terms in LSP text. A comparable corpus built for this research contains 167 English texts and 229 Chinese texts with around 600,000 English tokens and 900,000 Chinese characters. The corpus is annotated with mega-information and tagged with POS for further use. To get the key word list from the corpus, BFSU PowerConc software is used with the referential corpora of Crown and CLOB for English and TORCH and LCMC for Chinese. A VB program is written to generate the multi-word units, and then GOOGLE translators’ toolkit is used to get translation pairs and SDL trados fuzzy match function is applied to extract lists of multi-word terms and their translations. The results show this method has 70% of translated term pairs scoring 2.0 in a 0~3 grading scale with a 0.5 interval by human graders. The methods can be applied to extract translation term pairs for computer-aided translation of language for specific purpose texts. Also, the by-product comparable corpus, combined with N-gram multiword unit lists, can be used in facilitating trainee translators in translation. The findings underline the significance of combing the use of machine translation method with corpora techniques, and also foresee the necessity of comparable corpora building and sharing and Conc-gram extracting in this field.

Knowledge-poor and Knowledge-rich Approaches for Multilingual Terminology Extraction

In this paper, we present two terminology extraction tools in order to compare a knowledge-poor and a knowledge-rich approach. Both tools process single and multi-word terms and are designed to handle multilingualism. We run an evaluation on six languages and two di erent domains using crawled comparable corpora and hand-crafted reference term lists. We discuss the three main results achieved for terminology extraction. The rst two evaluation scenarios concern the knowledge-rich framework. Firstly, we compare performances for each of the languages depending on the ranking that is applied: speci city score vs. the number of occurrences. Secondly, we examine the relevancy of the term variant identi cation to increase the precision ranking for any of the languages. The third evaluation scenario compares both tools and demonstrates that a probabilistic term extraction approach, developed with minimal e ort, achieves satisfactory results when compared to a rule-based method.

Terminologija odnosov z javnostmi: korpus — luščenje — terminološka podatkovna zbirka

V prispevku prikazujemo analizo luščenja eno- in večbesednih terminoloških kandidatov, ki smo ga izvedli za potrebe priprave terminološke podatkovne zbirke odnosov z javnostmi na podlagi korpusa KoRP z luščilnikom LUIZ. Podrobneje se posvečamo dvojemu: (a) izluščenim enobesednim samostalniškim terminološkim kandidatom, katerih seznam primerjamo s pogostostnim seznamom samostalnikov v korpusu KoRP in vrednotimo glede na terminološkost, kot sta jo prepoznala dva področna strokovnjaka, ter (b) izluščenim večbesednim nizom z glagolskim in samostalniškim jedrom. Nadgrajeno metodo luščenja in izboljšan prikaz rezultatov smo dopolnili še z analizo priklica. Potrdili oz. ugotovili smo, da je v primerjavi s pogostostnim seznamom terminološki potencial enot v zgornjem delu seznama izluščenih samostalnikov večji, da imajo izluščeni glagolski besedni nizi predvsem kolokacijsko vrednost, ne pa tudi terminološke, in da so najbolj terminološko produktivni vzorci luščenja samostalniških zvez z naslednjo zgradbo: [pridevnik + samostalnik], [pridevnik + in + pridevnik + samostalnik] ter [pridevnik + pridevnik + samostalnik]. Analiza priklica je pokazala predvsem nizko stopnjo strinjanja med obema področnima strokovnjakoma, sicer pa je bil priklic razmeroma visok.

Slovenian legal language from the standpoint of a Slovenian language expert

The paper draws attention to the current issues of legal language use within oneʼs own language, in this case Slovenian. We use specific cases to point out the additional semantic stress on legal terminology, word-formational variations and their (non-)acceptability, the need to take into consideration grammatical categories such as semantic definiteness in adjectives, and perfective and imperfective aspects in verbs and gerunds, as well as the (non-)justification of conversions within multiword legal terms. In this regard it is established that whatever is ambiguous in legal terminology is at the same time also wrong. Comment is made on the lexico-grammatical response of legal Slovenian to the inclusion of concepts of the transnational legal order, i.e. the legal order of the European Union. The analysis uses Slovenian legal texts and translated legal texts all approved by the competent legal institutions for linguistic analysis and the legal lexis which had already been included in language-oriented discussions. According to the issues presented, the study of law (i.e. legal norms, legal science and legal practice) reveals a constant need, or rather impetus, to introduce a parallel study course dealing with regular language based and linguistic updates of legal Slovenian.

Distributional Semantic Models for Affective Text Analysis

We present an affective text analysis model that can directly estimate and combine affective ratings of multi-word terms, with application to the problem of sentence polarity/semantic orientation detection. Starting from a hierarchical compositional method for generating sentence ratings, we expand the model by adding multi-word terms that can capture non-compositional semantics. The method operates similarly to a bigram language model, using bigram terms or backing off to unigrams based on a (degree of) compositionality criterion. The affective ratings for n-gram terms of different orders are estimated via a corpus-based method using distributional semantic similarity metrics between unseen words and a set of seed words. N-gram ratings are then combined into sentence ratings via simple algebraic formulas. The proposed framework produces state-of-the-art results for word-level tasks in English and German and the sentence-level news headlines classification SemEval'07-Task14 task. The inclusion of bigram terms to the model provides significant performance improvement, even if no term selection is applied.

Term extraction from sparse, ungrammatical domain-specific documents

Existing term extraction systems have predominantly targeted large and well-written document collections, which provide reliable statistical and linguistic evidence to support term extraction. In this article, we address the term extraction challenges posed by sparse, ungrammatical texts with domain-specific contents, such as customer complaint emails and engineers’ repair notes. To this aim, we present ExtTerm, a novel term extraction system. Specifically, as our core innovations, we accurately detect rare (low frequency) terms, overcoming the issue of data sparsity. These rare terms may denote critical events, but they are often missed by extant TE systems. ExtTerm also precisely detects multi-word terms of arbitrarily lengths, e.g. with more than 2 words. This is achieved by exploiting fundamental theoretical notions underlying term formation, and by developing a technique to compute the collocation strength between any number of words. Thus, we address the limitation of existing TE systems, which are primarily designed to identify terms with 2 words. Furthermore, we show that open-domain (general) resources, such as Wikipedia, can be exploited to support domain-specific term extraction. Thus, they can be used to compensate for the unavailability of domain-specific knowledge resources. Our experimental evaluations reveal that ExtTerm outperforms a state-of-the-art baseline in extracting terms from a domain-specific, sparse and ungrammatical real-life text collection.

Comparing social tags with subject headings on annotating books: A study comparing the information science domain in English and Chinese

The literature often views the emergence of social tagging as a potential alternative method to controlled vocabulary for organizing and indexing large-scale information resources. In this paper, we present an in-depth examination of the relationship between social tagging and controlled vocabulary-based indexing and organization in two unique contexts: the information science domain and when comparing data gathered from both English and Chinese sources. Our results show that the information science domain has more overlap between social tags and controlled vocabulary-based subject terms. This is reflected in the higher percentage of overlapping terms between tags and subject terms, as well as in the strong similarity (measured by Jaccard’s coefficient) in frequently used keywords among tags and subject terms. However, social tags in the information science domain still possess limitations in terms of uncontrolled terms, where inconsistencies and noisy usages exist. Our results also show that language difference does have an impact on social tagging. The numbers of Chinese tags overall and per book are less than those of English tags. The most frequently used English tags are single-word terms, which are different from multi-word controlled vocabulary terms. In comparison, the character difference between the most frequently used Chinese tags and Chinese subject terms is just one character (3 vs 4). However, English and Chinese users do share many similar behaviours when they tag books in the information science domain. Many of the most frequently used tags are shared between the two languages and the patterns of overlap between topical tags and subject terms are also similar between the two languages. Overall, despite the application limitations for social tagging in cataloguing and indexing, we believe that tagging has the potential to become a complementary resource for expanding and enriching controlled vocabulary systems. With the help of future technology to regulate and promote features related to controlled vocabulary in social tags, a hybrid cataloguing and indexing system that integrates social tags with controlled vocabulary would greatly improve people’s organizational and access capabilities within information resources.

First Token Algorithm for Searching Compound Terms Using Thesaurus Database

Problem statement: Searching text materials is the one of the most im portant operations that carried out by search engines either on web or desktop applications, searching algorithms are required sometimes to find a specific word into a t ext, others to find a multi word term (pattern matching) into a text. Searching for term into a th esaurus database can be carried out using many searching algorithm such as brute-force algorithm a nd others. Approach: We addressed several issues concerning developing a searching algorithm that se arch terms into thesaurus database. Two exact algorithms were discussed and compared. The first a lgorithm, brute-force algorithm and the second one were proposed by this study to enhance brute-force algorithm. Results: We proposed an efficient search algorithm and compare it with brute force technique . Computational results showed that our algorithm can provide an efficient search algorithm that redu ces the number of queries and the total time requir ed to finish the required task. Conclusion: Our study showed an optimum solution for larger si ze of the studied problem with much less processing time than the brute-force algorithm. The modified algorithm has a higher efficiency to deal with Thes aurus Database searching problems.

Learning morphosyntactic patterns for multiword term extraction

The identification of valid terms in any domain is fundamental to its computerization. For this reason, in this paper we present a method for obtaining automated morphosyntactic patterns, which will help researchers obtain valid terms from the proposed patterns, in order to build quality ontologies for the translation from one language to another, or to find relevant terms in short sentences, which can be used as parameters in question-answer systems. For this purpose, we use some statistical methods which show candidates in a pattern vector. Then, a heuristic process unfolds to refine the pattern vector obtained, basing on two main parameters: the statistical results previously obtained and the length of the pattern analyzed. As a result, we obtain the collection of the best patterns for the detection of real multiword terms. Key words: Morphosyntactic patterns, multiword terms, incremental learning.

Identifying patterns for unsupervised learning of multiword terms

The identification of valid terms in any domain is fundamental to its computerization. For this reason, in this paper we present a method for obtaining automated morphosyntactic patterns, which will help researchers to obtain valid terms from the proposed patterns, in order to build quality ontologies for the translation from one language to another, or to find important concepts in short sentences, which can be used as parameters in question-answer systems. For this purpose, we use some statistical methods which show candidates in a pattern vector. Then, a heuristic process unfolds to refine the pattern vector obtained, based on two main parameters: the statistical results previously obtained and the pattern length analyzed. As a result, we obtain the collection of the best patterns for the detection of real multiword terms. Key words: Morphosyntactic patterns, multiword terms, incremental learning.

Revealing Interdisciplinarity in Nanoscience and Technology Queries: A Transaction Log Analysis Approach

KO KNOWLEDGE ORGANIZATION , Seite 135 - 153

Willow: a system to automatically assess students' free-text answers by using a combination of shallow NLP techniques

In this paper, a new approach to automatically assessing students’ short free-text answers is presented. It has been implemented into the multilingual web-based application called Willow. Willow is able to formulate a question (chosen randomly or according to the students’ personal features) and, to score the answer, written in Spanish or in English, by using a combination of shallow NLP techniques such as stemming, removal of closed-class words, word sense disambiguation or multiword term identification. Several experiments have been carried out to identify the best combination of techniques for each language to find the optimal configuration in each case.