Landauer, Th. K. and Littman, M. L. A statistical method for language-independent representation of the topical content of text segments. Proceedings of the Sixth Annual Conference of the UW Centre for the New Oxford English Dictionary and Text Research, 1990.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....preprocessing is necessary to get the necessary information such as the mentioned lexical associations. 2.2. Automatic, Unrestricted Indexing Traditional approaches to IR as well as to CLIR are based on surface word forms and their statistical distribution such as Latent Semantic Indexing (LSI) [5]. These approaches in practice are still the most successful ones, though they ignore the fact that terms may be ambiguous. Neglecting lexical ambiguity reduces the performance of IR as non relevant documents are retrieved on the basis of the irrelevant readings of terms. Separating relevant ....

Landauer, Th. K. and Littman, M. L. A statistical method for language-independent representation of the topical content of text segments. Proceedings of the Sixth Annual Conference of the UW Centre for the New Oxford English Dictionary and Text Research, 1990.

....try to get advantage from both approaches. The last ones can be further categorised by the type of resources they use, i.e. machine translation systems (Gachot et al., 1998) transfer dictionaries (Gey et al., 1999) or corpus based resources deployed, for instance, by using latent semantic indexing (Landauer Littman, 1990) or similarity thesauri (Sheridan et al., 1997) The use of linguistic knowledge provided by techniques from natural language processing is proven to be most successful to increase the effectiveness of a retrieval system (Liddy, 1998) If only stemming information for indexing and query processing ....

Landauer, T.K.& Littman,M.L. (1990). A statistical method for language-independent representation of the topical content of text segments. In Proceedings of the Sixth Annual Conference of the UW Centre for the New Oxford English Dictionary and Text Research.

....try to get advantage from both approaches. The last ones can be further categorised by the type of resources they use, i.e. machine translation systems (Gachot et al., 1998) transfer dictionaries (Gey et al., 1999) or corpus based resources, deployed, for instance, by using latent semantic indexing (Landauer Littman, 1990) or similarity thesauri (Sheridan et al., 1997) The use of linguistic knowledge provided by techniques from natural language processing is proven to be most successful to increase the effectiveness of a retrieval system (Liddy, 1998) If only stemming information for indexing and query processing ....

Landauer, T.K.& Littman,M.L. (1990). A statistical method for language-independent representation of the topical content of text segments. In Proceedings of the Sixth Annual Conference of the UW Centre for the New Oxford English Dictionary and Text Research.

....because it reconstructs large scale index data for existing IR engines. Finally, interlingual representation approach projects both queries and documents into languageindependent representation, such as (bilingual) thesaurus classes [8, 10, 14, 20, 22] and languageindependent vector space models [4, 7, 13]. However, these methods generally require (manual) alignment of bilingual thesauri corpora, and are thus expensive. In addition, Carbonell et al. 4] showed that the corpus based query translation method outperforms language independent vector space models. To sum up, we prefer the query ....

Thomas K. Landauer and Michael L. Littman. A statistical method for language-independent representation of the topical content of text segments. In Proceedings of the 11th International Conference: Expert Systems and Their Applications, pp. 77--85, 1991.

....one language. Automatic language identification therefore becomes a critical part of the indexing task. 3 Language Identification Although there are several approaches to information retrieval which may be considered to be completely language independent (Latent Semantic Indexing for example, [Landauer Littman, 1991]) most approaches employ at least some form of lexical analysis, like stemming as outlined above, or even more advanced linguistic analyses of the texts. Since these analyses are in general language specific, it becomes important to be able to reliably identify the language of a document to be ....

Landauer, T. K., & Littman, M. L. (1991). A Statistical Method for Language-Independent Representation of the Topical Content of Text Segments. In Proceedings of the Eleventh International Conference: Expert Systems and Their Applications, pp. 77--85.

....analysis. Because no human effort is required for knowledge engineering, the LSI method is widely applicable to different domains. The LSI method can be used both for information retrieval and information filtering applications [5] It has also been used successfully for crosslanguage retrieval [6], and as a way of suggesting new indexing terms for conventional retrieval methods. Several experiments using LSI for information filtering will be summarized. An important issue in this work is how best to represent people s interests e.g. the words in their self described interests; ....

Thomas K. Landauer and Michael L. Littman. A statistical method for language-independent representation of the topical content of text segments. In Proceedings of the Eleventh International Conference: Expert Systems and Their Applications, volume 8, pages 77--85. Avignon France, May 1991.

....learned from that research that would be useful for the applications presently being investigated. In 1990, Landauer and Littman (then with Bellcore) developed a corpus based cross language free text retrieval technique which has come to be known as Cross Language Latent Semantic Indexing (CLLSI) [17, 18]. The remarkable thing about this work is that in addition to beginning the present development of cross language text retrieval systems in the USA, it remains to this day the only technique that has demonstrated cross language text retrieval effectiveness that is on a par with the within language ....

Thomas K. Landauer and Michael L. Littman. A statistical method for language-independent representation of the topical content of text segments. In Proceedings of the Eleventh International Conference: Expert Systems and Their Applications, volume 8, pages 77--85, Avignon France, May 1991.

....an explicit query. and suppress the effect of varying term usage. Documents can then be compared using the cosine similarity measure and ranked for display in the usual way. LSI has been applied to multilingual text retrieval in a similar way to the relevance feedback technique described above [5, 37, 38, 76]. The basic approach is best illustrated by Landauer and Littman [38] Randomly selecting 900 training paragraphs and 1,582 evaluation paragraphs from the Hansards collection, a parallel corpus of Canadian parliamentary proceedings, they first applied LSI to identify the principal components of ....

....can then be compared using the cosine similarity measure and ranked for display in the usual way. LSI has been applied to multilingual text retrieval in a similar way to the relevance feedback technique described above [5, 37, 38, 76] The basic approach is best illustrated by Landauer and Littman [38]. Randomly selecting 900 training paragraphs and 1,582 evaluation paragraphs from the Hansards collection, a parallel corpus of Canadian parliamentary proceedings, they first applied LSI to identify the principal components of the training set. When LSI is applied to a parallel corpus, the matrix ....

Thomas K. Landauer and Michael L. Littman. A statistical method for languageindependent representation of the topical content of text segments. In Proceedings of the Eleventh International Conference: Expert Systems and Their Applications, volume 8, pages 77--85. Avignon France, May 1991.

No context found.

Thomas K. Landauer and Michael L. Littman. A statistical method for language-independent representation of the topical content of text segments. In Proceedings of the Eleventh International Conference: Expert Systems and Their Applications, volume 8, pages 77--85, Avignon France, May 1991.

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