| Article ID | Journal | Published Year | Pages | File Type |
|---|---|---|---|---|
| 38564 | World Patent Information | 2008 | 6 Pages |
Abstract
Latent semantic indexing (LSI) can be used in patent searching to overcome drawbacks of Boolean searching and to give more accurate retrieval. LSI combines the vector space model (VSM) of document retrieval with single value decomposition (SVD), using linear algebra techniques to uncover word relationships in the text. Results can be enhanced by using text clustering and tailoring SVD parameters to the specific corpus, in this case, patents, and by employing techniques to address ambiguities in language.
Related Topics
Physical Sciences and Engineering
Chemical Engineering
Bioengineering
Authors
James F. Ryley, Jeff Saffer, Andy Gibbs,