Stephen I. Gallant. A practical approach for representing context and for performing word sense disambiguation using neural networks. Neural Computation, 3(3):293--309, 1991.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....The vectors of relevant documents are made more similar to the respective queries by adding query terms to the documents. This method is an adaptive approach to IR, so we will defer further discussion until the section on adaptive methods later in this chapter. The approach used by Gallant [51] [50] is to represent each term in a collection using a multidimensional feature vector. Dimensions of the vector correspond to semantic features, such as human or light, engineered by the system designer. Concepts are also represented in this feature space, so a term such as star (in the sky) ....

....the potential to overcome this limitation. The feature that unifies these methods is that they each represent documents in a vector space such that two document vectors may be similar even if they share no terms. These methods include Latent Semantic Indexing [32] Gallant s context vector method [51], Brauen s adaptive document vectors [15] and Yang Chute s canonical concept mapping [166] 165] most of which were reviewed in Chapter 2. 3.1.2 Metric Similarity Modeling The approach we propose, Metric Similarity Modeling (MSM) also uses a multidimensional semantic space in which to ....

Stephen I. Gallant. A practical approach for representing context and for performing word sense disambiguation using neural networks. Neural Computation, 3(3):293--309, 1991.

....change frequently, and 33 ffl efficiently comparing queries to documents (i.e. finding near neighbors in high dimension spaces) 5.7 Related Work A number of other researchers are using related linear algebra methods for information retrieval and classification work. Schutze [27] and Gallant [14] have used SVD and related dimension reduction ideas for word sense disambiguation and information retrieval work. Hull [17] and Yang and Chute [29] have used LSI SVD as the first step in conjunction with statistical classification (e.g. discriminant analysis) Using the LSI derived dimensions ....

S. I. GALLANT, A practical approach for representing contexts and for performing word sense disambiguation using neural networks, Neural Computation, 3 (1991), pp. 293--309.

....of neurons in the brain. Whether such groups can serve as the basic units of information processing would need to be confirmed experimentally. A number of other connectionist models of lexical access and lexical disambiguation have been proposed recently (Bookman 1989; Cottrell and Small 1983; Gallant 1991; Gasser 1988; Gigley 1988; Kawamoto 1988; Sharkey 1989; Small 1990; Waltz and Pollack 1985) These models aim at explaining lexical processing with low level mechanisms, focusing on the timing of the process as well as on certain types of performance errors and deficits. They are primarily ....

Gallant, S. I. 1991. A practical approach for representing context and for performing word sense disambiguation using neural networks. Neural Computation, 3:293--309.

....by some appropriate vector metric) The experiment described below explores this idea by comparing meaning vectors which are close to each other to synonyms from an actual thesaurus. PREVIOUS WORK Extracting meaning vectors from large corpora is a relatively new but well explored area. Gallant [Gallant, 1991] proposed the idea of context vectors to represent meaning within a certain context, but his proposal involved hand coding the vectors, a laborious task. Both Sch utze [Sch utze, 1992] Sch utze, 1993] and Hecht Nielsen [Hecht Nielsen, 1994] proposed automatic techniques based directly on word ....

Gallant, S. (1991). A practical approach for representing context and for performing word sense disambiguation using neural networks. Neural Computation, 3:293--309.

....eaten etc. Therefore one would not expect to gain much by including more local syntactic information. Another question is whether features of lexical content and situational context would improve performance. It is difficult to obtain reliable features describing situational context (but cf. [Gallant, 1991] for an attempt) Lexical content features would have to be imported from other analyzing modules, a problem that has not been satisfactorily solved (cf. McClelland et al. 1989] Scheler, 1989] for a proposal on lexical syntactic interaction during processing) An analysis of the example ....

S.I. Gallant. A practical approach for representing context and for performing word sense disambiguation using neural networks. Neural Computation 3(3), pages 293--309, 1991.

....eaten etc. Therefore one would not expect to gain much by including more local syntactic information. Another question is whether features of lexical content and situational context would improve performance. It is difficult to obtain reliable features describing situational context (but cf. [Gallant, 1991] for an attempt) Lexical content features would have to be imported from other analyzing modules, a problem that has not been satisfactorily solved (cf. McClelland et al. 1989] Scheler, 1989] for a proposal on lexical syntactic interaction during processing) An analysis of the example ....

S.I. Gallant. A practical approach for representing context and for performing word sense disambiguation using neural networks. Neural Computation, 3:293--309, 1991.

.... models rely either on hand crafted vector representations of the meanings of words (Kawamoto, 1993; Plaut Shallice, 1993) or randomly generated vector representations (Cottrell Plunkett, 1995; Plaut, 1995) Hand crafted representations may impose some structure on the fields of the vector (Gallant, 1991; Plaut Shallice, 1993) in an attempt to provide consistency, or they may be built ad hoc (Kawamoto, 1993) These techniques are motivated by the desire to relate the lexical representations used in the simulation to real words, so that parallels may be drawn to psycholinguistic experiments, and ....

Gallant, S.I. (1991). A practical approach for representing context and for performing word sense disambiguation using neural networks. Neural Computation, 3.

....change frequently, and ffl efficiently comparing queries to documents (i.e. finding near neighbors in high dimension spaces) 5.7. Related Work. A number of other researchers are using related linear algebra methods for information retrieval and classification work. Schutze [26] and Gallant [13] have used SVD and related dimension reduction ideas for word sense disambiguation and information retrieval work. Hull [16] and Yang and Chute [28] have used LSI SVD as the first step in conjunction with statistical classification (e.g. discriminant analysis) Using the LSI derived dimensions ....

S. I. Gallant, A practical approach for representing contexts and for performing word sense disambiguation using neural networks, Neural Computation, 3 (1991), pp. 293--309.

No context found.

-240. #29# Stephen I. Gallant. A practical approach for representing context and for performing word sense disambiguation using neural nets. Neural Computation,

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