Search (5 results, page 1 of 1)

0.0026742492 = product of:
  0.0053484985 = sum of:
    0.0053484985 = product of:
      0.010696997 = sum of:
        0.010696997 = weight(_text_:a in 1910) [ClassicSimilarity], result of:
          0.010696997 = score(doc=1910,freq=20.0), product of:
            0.053105544 = queryWeight, product of:
              1.153047 = idf(docFreq=37942, maxDocs=44218)
              0.046056706 = queryNorm
            0.20142901 = fieldWeight in 1910, product of:
              4.472136 = tf(freq=20.0), with freq of:
                20.0 = termFreq=20.0
              1.153047 = idf(docFreq=37942, maxDocs=44218)
              0.0390625 = fieldNorm(doc=1910)
      0.5 = coord(1/2)
  0.5 = coord(1/2)

Abstract

Purpose - The purpose of this article is to examine the effectiveness of the unified medical language system (UMLS) semantic network as a seed ontology for building a medical field ontology. Design/methodology/approach - The information extraction process known as the knowledge engineering approach was used to extract concepts and their semantic relations from documents on "colon cancer treatment". The UMLS semantic network was used as a seed ontology, and was extended and enriched with the extracted concepts and semantic relations using Protégé. Findings - Only half of the semantic relations extracted manually were defined (or inferable) in the UMLS semantic network. The remaining half could be added to the network to extend its coverage. In addition, two semantic types in the network were found to be too general and four new sublevel semantic types were proposed to make them more specific. Research limitations/implications - Since only 109 research paper abstracts in the "colon cancer treatment" domain were analyzed in this study, more abstracts from the colon cancer treatment domain as well as from other cancer treatment domains (such as breast cancer treatment) can be analyzed to give a better generalization of our findings. Originality/value - This study shares our findings on the effectiveness of the UMLS semantic network as a seed ontology for building a medical domain ontology, and also provides the basic guidelines for building or extending a medical domain ontology using the UMLS.

Type

a

0.0023919214 = product of:
  0.0047838427 = sum of:
    0.0047838427 = product of:
      0.009567685 = sum of:
        0.009567685 = weight(_text_:a in 2661) [ClassicSimilarity], result of:
          0.009567685 = score(doc=2661,freq=16.0), product of:
            0.053105544 = queryWeight, product of:
              1.153047 = idf(docFreq=37942, maxDocs=44218)
              0.046056706 = queryNorm
            0.18016359 = fieldWeight in 2661, product of:
              4.0 = tf(freq=16.0), with freq of:
                16.0 = termFreq=16.0
              1.153047 = idf(docFreq=37942, maxDocs=44218)
              0.0390625 = fieldNorm(doc=2661)
      0.5 = coord(1/2)
  0.5 = coord(1/2)

Abstract

This study is part of a project to develop an automatic method to build ontologies, especially in a medical domain, from a document collection. An earlier study had investigated an approach to inferring semantic relations between medical concepts using the UMLS (Unified Medical Language System) semantic net. The study found that semantic relations between concepts could be inferred 68% of the time, although the method often could not distinguish between a few possible relation types. Our current research focuses an the use of natural language processing techniques to improve the identification of semantic relations. In particular, we explore both a semi-automatic and manual construction of linguistic patterns for identifying treatment relations in medical abstracts in the domain of colon cancer treatment. Association rule mining was applied to sample sentences containing both a disease concept and a reference to drug, to identify frequently occurring word pattems to see if these pattems could be used to identify treatment relations in sentences. This did not yield many useful patterns, suggesting that statistical association measures have to be complemented with syntactic and semantic constraints to identify useful patterns. In the second part of the study, linguistic patterns were manually constructed based an the same sentences. This yielded promising results. Work is ongoing to improve the manually constructed pattems as well as to identify the syntactic and semantic constraints that can be used to improve the automatic construction of linguistic patterns.

Type

a

0.001757696 = product of:
  0.003515392 = sum of:
    0.003515392 = product of:
      0.007030784 = sum of:
        0.007030784 = weight(_text_:a in 1978) [ClassicSimilarity], result of:
          0.007030784 = score(doc=1978,freq=24.0), product of:
            0.053105544 = queryWeight, product of:
              1.153047 = idf(docFreq=37942, maxDocs=44218)
              0.046056706 = queryNorm
            0.13239266 = fieldWeight in 1978, product of:
              4.8989797 = tf(freq=24.0), with freq of:
                24.0 = termFreq=24.0
              1.153047 = idf(docFreq=37942, maxDocs=44218)
              0.0234375 = fieldNorm(doc=1978)
      0.5 = coord(1/2)
  0.5 = coord(1/2)

Abstract

This chapter examines the nature of semantic relations and their main applications in information science. The nature and types of semantic relations are discussed from the perspectives of linguistics and psychology. An overview of the semantic relations used in knowledge structures such as thesauri and ontologies is provided, as well as the main techniques used in the automatic extraction of semantic relations from text. The chapter then reviews the use of semantic relations in information extraction, information retrieval, question-answering, and automatic text summarization applications. Concepts and relations are the foundation of knowledge and thought. When we look at the world, we perceive not a mass of colors but objects to which we automatically assign category labels. Our perceptual system automatically segments the world into concepts and categories. Concepts are the building blocks of knowledge; relations act as the cement that links concepts into knowledge structures. We spend much of our lives identifying regular associations and relations between objects, events, and processes so that the world has an understandable structure and predictability. Our lives and work depend on the accuracy and richness of this knowledge structure and its web of relations. Relations are needed for reasoning and inferencing. Chaffin and Herrmann (1988b, p. 290) noted that "relations between ideas have long been viewed as basic to thought, language, comprehension, and memory." Aristotle's Metaphysics (Aristotle, 1961; McKeon, expounded on several types of relations. The majority of the 30 entries in a section of the Metaphysics known today as the Philosophical Lexicon referred to relations and attributes, including cause, part-whole, same and opposite, quality (i.e., attribute) and kind-of, and defined different types of each relation. Hume (1955) pointed out that there is a connection between successive ideas in our minds, even in our dreams, and that the introduction of an idea in our mind automatically recalls an associated idea. He argued that all the objects of human reasoning are divided into relations of ideas and matters of fact and that factual reasoning is founded on the cause-effect relation. His Treatise of Human Nature identified seven kinds of relations: resemblance, identity, relations of time and place, proportion in quantity or number, degrees in quality, contrariety, and causation. Mill (1974, pp. 989-1004) discoursed on several types of relations, claiming that all things are either feelings, substances, or attributes, and that attributes can be a quality (which belongs to one object) or a relation to other objects.
Linguists in the structuralist tradition (e.g., Lyons, 1977; Saussure, 1959) have asserted that concepts cannot be defined on their own but only in relation to other concepts. Semantic relations appear to reflect a logical structure in the fundamental nature of thought (Caplan & Herrmann, 1993). Green, Bean, and Myaeng (2002) noted that semantic relations play a critical role in how we represent knowledge psychologically, linguistically, and computationally, and that many systems of knowledge representation start with a basic distinction between entities and relations. Green (2001, p. 3) said that "relationships are involved as we combine simple entities to form more complex entities, as we compare entities, as we group entities, as one entity performs a process on another entity, and so forth. Indeed, many things that we might initially regard as basic and elemental are revealed upon further examination to involve internal structure, or in other words, internal relationships." Concepts and relations are often expressed in language and text. Language is used not just for communicating concepts and relations, but also for representing, storing, and reasoning with concepts and relations. We shall examine the nature of semantic relations from a linguistic and psychological perspective, with an emphasis on relations expressed in text. The usefulness of semantic relations in information science, especially in ontology construction, information extraction, information retrieval, question-answering, and text summarization is discussed. Research and development in information science have focused on concepts and terms, but the focus will increasingly shift to the identification, processing, and management of relations to achieve greater effectiveness and refinement in information science techniques. Previous chapters in ARIST on natural language processing (Chowdhury, 2003), text mining (Trybula, 1999), information retrieval and the philosophy of language (Blair, 2003), and query expansion (Efthimiadis, 1996) provide a background for this discussion, as semantic relations are an important part of these applications.

Type

a

0.0016913437 = product of:
  0.0033826875 = sum of:
    0.0033826875 = product of:
      0.006765375 = sum of:
        0.006765375 = weight(_text_:a in 2624) [ClassicSimilarity], result of:
          0.006765375 = score(doc=2624,freq=8.0), product of:
            0.053105544 = queryWeight, product of:
              1.153047 = idf(docFreq=37942, maxDocs=44218)
              0.046056706 = queryNorm
            0.12739488 = fieldWeight in 2624, product of:
              2.828427 = tf(freq=8.0), with freq of:
                8.0 = termFreq=8.0
              1.153047 = idf(docFreq=37942, maxDocs=44218)
              0.0390625 = fieldNorm(doc=2624)
      0.5 = coord(1/2)
  0.5 = coord(1/2)

Abstract

This paper reports a study in automatic sentiment classification, i.e., automatically classifying documents as expressing positive or negative Sentiments/opinions. The study investigates the effectiveness of using SVM (Support Vector Machine) an various text features to classify product reviews into recommended (positive Sentiment) and not recommended (negative sentiment). Compared with traditional topical classification, it was hypothesized that syntactic and semantic processing of text would be more important for sentiment classification. In the first part of this study, several different approaches, unigrams (individual words), selected words (such as verb, adjective, and adverb), and words labelled with part-of-speech tags were investigated. A sample of 1,800 various product reviews was retrieved from Review Centre (www.reviewcentre.com) for the study. 1,200 reviews were used for training, and 600 for testing. Using SVM, the baseline unigram approach obtained an accuracy rate of around 76%. The use of selected words obtained a marginally better result of 77.33%. Error analysis suggests various approaches for improving classification accuracy: use of negation phrase, making inference from superficial words, and solving the problem of comments an parts. The second part of the study that is in progress investigates the use of negation phrase through simple linguistic processing to improve classification accuracy. This approach increased the accuracy rate up to 79.33%.

Type

a

0.0014647468 = product of:
  0.0029294936 = sum of:
    0.0029294936 = product of:
      0.005858987 = sum of:
        0.005858987 = weight(_text_:a in 2125) [ClassicSimilarity], result of:
          0.005858987 = score(doc=2125,freq=6.0), product of:
            0.053105544 = queryWeight, product of:
              1.153047 = idf(docFreq=37942, maxDocs=44218)
              0.046056706 = queryNorm
            0.11032722 = fieldWeight in 2125, product of:
              2.4494898 = tf(freq=6.0), with freq of:
                6.0 = termFreq=6.0
              1.153047 = idf(docFreq=37942, maxDocs=44218)
              0.0390625 = fieldNorm(doc=2125)
      0.5 = coord(1/2)
  0.5 = coord(1/2)

Abstract

Blogs are readily available sources of opinions and sentiments that in turn could influence the opinions of the blog readers. Previous studies have attempted to infer influence from blog features, but they have ignored the possible influence styles that describe the different ways in which influence is exerted. We propose a novel approach to analyzing bloggers' influence styles and using the influence styles as features to improve the performance of influence diffusion detection among linked bloggers. The proposed influence style (INFUSE) model describes bloggers' influence through their engagement style, persuasion style, and persona. Methods used include similarity analysis to detect the creating-sharing aspect of engagement style, subjectivity analysis to measure persuasion style, and sentiment analysis to identify persona style. We further extend the INFUSE model to detect influence diffusion among linked bloggers based on the bloggers' influence styles. The INFUSE model performed well with an average F1 score of 76% compared with the in-degree and sentiment-value baseline approaches. Previous studies have focused on the existence of influence among linked bloggers in detecting influence diffusion, but our INFUSE model is shown to provide a fine-grained description of the manner in which influence is diffused based on the bloggers' influence styles.

Type

a