Gregory Grefenstette. 1992. Finding semantic similarity in raw text: The Deese antonyms. In Probabilistic Approaches to Natural Language: Papers from the 1992 Fall Symposium. AAAI.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....specifically for cooccurrence information, we can take advantage of the huge document collections that have been indexed by modern Web search engines. Various measures of semantic similarity between word pairs have been proposed, some using statistical (unsupervised learning from text) techniques [16, 17, 18], some using lexical databases (hand built) 19, 20] and some hybrid approaches, combining statistics and lexical information [21, 22] Statistical techniques typically suffer from the sparse data problem: they perform poorly when the words are relatively rare, due to the scarcity of data. Hybrid ....

Grefenstette, G.: Finding Semantic Similarity in Raw Text: The Deese Antonyms. In: R. Goldman, P. Norvig, E. Charniak and B. Gale (eds.), Working Notes of the AAAI Fall Symposium on Probabilistic Approaches to Natural Language. AAAI Press (1992) 61-65.

....to discover the maximally informative binary question at a tree node. Fisher and Riloff [55] use the t statistic as a measure of co occurrence likelihood between two items. It too is calculated from corpus frequency information and can indicate strong 90 correlations between items. Grefenstette [70] suggests that the Jaccard distance similarity measure leads to interesting language collocations. Kneser and Ney [92] and Ney, Essen and Kneser [116] include examples of word classification systems which, while not hierarchically clustered, use an optimisation technique based on decision directed ....

Gregory Grefenstette. Finding semantic similarity in raw text : The deese antonyms. In Probabilistic Approaches to Natural Language. American Association for Artificial Intelligence, AAAI Press, 1992. Technical report FS-92-05.

....and can indicate strong correlations between items. This statistic can measure collocational differences, whereas mutual information measures collocational similarities [29] The likelihood ratio test allows measures of collocational similarity without assuming a normal distribution. Grefenstette [59] suggests that an adaptation of the Jaccard distance similarity measure leads to interesting language collocations for example, his measure can be used to discover some interesting antonyms. The Jaccard measure is similar to Brill et al. s distributional statistic; the measure is calculated ....

Gregory Grefenstette. Finding semantic similarity in raw text : The Deese antonyms. In Probabilistic Approaches to Natural Language. American Association for Artificial Intelligence, AAAI Press, 1992. Technical report FS-92-05.

....of the same syntactic class, but with less semantic homogeneity. In each example below, the words listed are the entire contents of the node mentioned. The most striking property of the clusters produced was the classification of words into coherent semantic fields. Grefenstette has pointed out [Grefenstette, 1992] that the Deese antonyms, such as large and small or hot and cold show up commonly in these analyses. Our methods discovered entire graded domains, rather than just pairs of opposites. The following sample node shows a set of seventeen adjectives describing comparative magnitudes: ....

Grefenstette, G. (1992). Finding Semantic Similarity in Raw Text: the Deese Antonyms. In AAAI Fall Symposium Series: Probabilistic Approaches to Natural Language (Working Notes), (pp. 61-66). Cambridge, MA.

....between words, has frequently drawn the interest of researchers in natural language processing. Many of these studies aimed at establishing the semantical similarity between words (Brown et al. 1992; Dagan, Markus, and Markovitch, 1993; Dagan, Pereira, and Lee, 1994; Pereira and Tishby, 1992; Grefenstette, 1992). We suggest a method for clustering words purely on the basis of syntactical behavior. We show how the necessary data for such clustering can easily be drawn from a publicly available treebank, and how distinct types of behavior can be discovered. Although a part of speech tag set can be thought ....

Grefenstette, G. 1992. Finding semantic similarity in raw text: the Deese antonyms. In Working Notes, Fall Symposium Series, AAAI, pages 61--68.

....true broad coverage NLP. Much more promising, in our view, are automated methods for building large computational lexicons. For instance, statistical methods can be used to acquire information about the syntactic and semantic properties of words from large corpora (e.g. Basili et al. 1992; Grefenstette, 1992). Other work attempts to extract semantic information from a (partial) analysis of free text (Grishman and Sterling, 1992; Hearst, 1992) Currently, however, none of these techniques appears capable of providing lexical information in sufficient detail. Another area of current interest is the ....

Grefenstette, G. 1992. Finding Semantic Similarity in Raw Text: the Deese Antonyms. In Proceedings of AAAI Fall Symposium Series: Probabilistic Approaches to Natural Language , October 2325, 1992, 61-65.

....from the limitations in floating point precision inherent in all architectures. 3 Design Considerations Contemporary research based on corpus derived statistics has shown significant promise with respect to tasks such as the extraction of syntactic and semantic information (cf. Grefenstette [5], and Schutze [6] However, at least one constraint on the practicality of such approaches being used in real world applications is the availability of the statistical information upon which they rely. If access to this sort of information is slow and inefficient, the ability to apply such ....

Gregory Grefenstette. Finding semantic similarity in raw text: The Deese antonyms. In Fall Symposium Series, Working Notes, Probabilistic Approaches to Natural Language, pages 61--65, Cambridge, MA, October 1992. American Association for Artificial Intelligence.

....Conference, University of Edinburgh, June 18 19, 1997. This research was partially supported by a Postgraduate Scholarship from the Natural Sciences and Engineering Research Council of Canada. SEXTANT was able to capture some of the accepted semantic similarity of common English words . (Grefenstette, 1992:61) we are assuming that words that behave in the same way on the surface have semantic similarities. Charniak, 1993:134) An important question remains: how valid is the semantic distance (or contextual similarity) measure obtained through statistical analysis of large corpora In order ....

Grefenstette, G. (1992). Finding semantic similarity in raw text: the Deese antonyms. In R Goldman, P. Norvig, E. Charniak and B. Gale (Eds.) Working Notes of the AAAI Fall Symposium on Probabilistic Approaches to Natural Language. AAAI Press.

No context found.

Gregory Grefenstette. 1992. Finding semantic similarity in raw text: The Deese antonyms. In Probabilistic Approaches to Natural Language: Papers from the 1992 Fall Symposium. AAAI.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC