Reducing Information Variation in Text

Savary, Agata; Jacquemin, Christian

doi:10.1007/978-3-540-45115-0_6

Cited by 14 publications

(8 citation statements)

References 82 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Certainly, this is not an easy task. On the one hand, snippets provide (a) localized contextual paragraphs that are highly related to the query, (b) these localized contextual paragraphs express ideas and concepts by means of different paraphrases, which consist primarily of morphological, semantical, orthographical and syntactical variations of these ideas and concepts (Savary and Jacquemin, 2000), which makes the identification of promising answer candidates easier. On the other hand, search engines insert intentional breaks in snippets, in order to show relations among words relevant to the query that are separated by a large span of text.…”

Section: How Are Inventor and Invention Related? What Is The Relationmentioning

confidence: 99%

Genetic Algorithms for Data-Driven Web Question Answering

Figueroa

Neumann

2008

Evolutionary Computation

View full text Add to dashboard Cite

We present an evolutionary approach for the computation of exact answers to natural languages (NL) questions. Answers are extracted directly from the N–best snippets, which have been identified by a standard Web search engine using NL questions. The core idea of our evolutionary approach to Web question answering is to search for those substrings in the snippets whose contexts are most similar to contexts of already known answers. This context model together with the words mentioned in the NL question are used to evaluate the fitness of answer candidates, which are actually randomly selected substrings from randomly selected sentences of the snippets. New answer candidates are then created by applying specialized operators for crossover and mutation, which either stretch and shrink the substring of an answer candidate or transpose the span to new sentences. Since we have no predefined notion of patterns, our context alignment methods are very dynamic and strictly data-driven. We assessed our system with seven different datasets of question/answer pairs. The results show that this approach is promising, especially when it deals with specific questions.

show abstract

Section: How Are Inventor and Invention Related? What Is The Relationmentioning

confidence: 99%

Genetic Algorithms for Data-Driven Web Question Answering

Figueroa

Neumann

2008

Evolutionary Computation

View full text Add to dashboard Cite

show abstract

“…Conflations. We define a conflation [2] to be a syntactic paraphrase of a reference to a concept. [knife grinding] can appear in a document as the conflation "grind the knives" and [filter cleaning] can be conflated to "filters are cleaned".…”

Section: Noise Reductionmentioning

confidence: 99%

Enhancing knowledge mapping using automatically derived concepts

Anjewierden

Huijsen

Grootveld

2005

Proceedings of the 3rd International Conference on Knowledge Capture

View full text Add to dashboard Cite

Knowledge-mapping tools enable users to quickly identify relevant information and expertise. This paper discusses a number of natural-language phenomena that limit the performance of a straightforward approach. An empirical study on a real-life community provides a quantitative indication of the impact of this noise on the markup of concepts and retrieval of documents. We then discuss enhancing the usefulness of knowledgemapping tools through automatically derived concepts. General TermsDesign

show abstract

“…VMWEs have been the focus of much attention, both in linguistics and in natural language processing [11,1,3,15]. From a linguistic point of view, they present restricted variability patterns, licensing phenomena such as passivization, pronominalization of components, reordering, and free PP-movement depending on the VMWE category [14,10,17,8]. Moreover, verbs (and VMWEs) tend to have rich morphological inflection paradigms, and allow many (but not all) syntactic changes [6,5].…”

Section: Introductionmentioning

confidence: 99%

A Corpus Study of Verbal Multiword Expressions in Brazilian Portuguese

Ramisch

Ramisch²,

Zilio

et al. 2018

Lecture Notes in Computer Science

View full text Add to dashboard Cite

Verbal multiword expressions (VMWEs) such as to make ends meet require special attention in NLP and linguistic research, and annotated corpora are valuable resources for studying them. Corpora annotated with VMWEs in several languages, including Brazilian Portuguese, were made freely available in the PARSEME shared task. The goal of this paper is to describe and analyze this corpus in terms of the characteristics of annotated VMWEs in Brazilian Portuguese. First, we summarize and exemplify the criteria used to annotate VMWEs. Then, we analyze their frequency, average length, discontinuities and variability. We further discuss challenging constructions and borderline cases. We believe that this analysis can improve the annotated corpus and its results can be used to develop systems for automatic VMWE identification.

show abstract

Reducing Information Variation in Text

Cited by 14 publications

References 82 publications

Genetic Algorithms for Data-Driven Web Question Answering

Genetic Algorithms for Data-Driven Web Question Answering

Enhancing knowledge mapping using automatically derived concepts

A Corpus Study of Verbal Multiword Expressions in Brazilian Portuguese

Contact Info

Product

Resources

About