Compound-term processing: Difference between revisions

Revision as of 11:14, 23 June 2008

Revision as of 11:52, 11 September 2008

Line 1:

'''Compound term processing''' is the name that is used for a category of techniques in [[Information retrieval]] applications that performs matching on the basis of compound terms. Compound terms are built by combining two (or more) simple terms, for example "triple" is a single word term but "triple heart bypass" is a compound term. For the purpose of information retrieval, it is important to decide which terms to use for searching. In other words, the tasktasks isare to decide (i) if the whole compound term should be considered as an index and, (ii) which of its components are the more relevant for information retrieval.

In August 2003 [[Concept Searching Limited]] introduced the idea of using statistical Compound Term Processing via an article published in INFORMATION MANAGEMENT AND TECHNOLOGY (VOL 36 PART 4). A British Library Direct catalogue entry can be found here: <ref>[http://direct.bl.uk/bld/PlaceOrder.do?UIN=138451913&ETOC=RN] British Library Direct catalogue entry</ref>.

Line 11:

Most search engines simply look for documents that contain the words that the user enters into the search box (aka "keyword search" engines). [[Boolean search]] engines add a degree of sophistication by allowing the user to specify additional requirements but most users struggle to comprehend and use the necessary syntax (e.g. Tiger NEAR Woods AND (golf OR golfing) NOT Volkswagen). [[Phrase search]] is easier to understand but can lead to many useful documents being missed if they do not contain the exact phrase specified.

Techniques for probabilistic weighting of single word terms dates back to at least 1976 and the landmark publication by Stephen Robertson and [[Karen Spärck Jones]]: Relevance weighting of search terms originally published in the Journal of the American Society for Information Science. <ref> [http://www.soi.city.ac.uk/~ser/papers/RSJ76.pdf] Relevance weighting of search terms</ref> Professor Stephen Robertson has stated repeatedly that the assumption of word independence is “patently not justified”justified and exists simply as a “mattermatter of mathematical convenience”convenience. The objection to assumptions about term independence are not new, dating back to at least 1964 when H. H. Williams expressed it this way: "The assumption of independence of words in a document is usually made as a matter of mathematical convenience". <ref>WILLIAMS, J.H., 'Results of classifying documents with multiple discriminant functions', In : Statistical Association Methods for Mechanized Documentation, National Bureau of Standards, Washington, 217-224 (1965).</ref>

[[Compound Termterm Processingprocessing]] is a new approach to an old problem: how to improve the relevance of search results without missing anything important whilst maintaining ease of use. By forming compound (i.e. multi-word) terms and placing these in the search engine's index the search can be performed with a higher degree of accuracy because the ambiguity inherent in single words is no longer a problem. A search for ''survival rates following a triple heart bypass in elderly people'' will locate documents about this topic even if this precise phrase is not contained in any document. A [[concept search]] using "Compound Term Processing" can extract the key concepts automatically (in this case "survival rates" and, "triple heart bypass" and "elderly people") and use these to select the most relevant documents.

In 2004 Anna Lynn Patterson filed a number of patents on the subject of "Phrase based indexing and retrieval" and to which Google subsequently acquired the rights. A full discussion of the patents can be found here: [http://www.webmasterwoman.com/search-engines/phrase-based-indexing.html Webmaster Woman]. The patents themselves can be found online, for example: <ref>[http://appft1.uspto.gov/netacgi/nph-Parser?Sect1=PTO1&Sect2=HITOFF&d=PG01&p=1&u=%2Fnetahtml%2FPTO%2Fsrchnum.html&r=1&f=G&l=50&s1=%2220060031195%22.PGNR.&OS=DN/20060031195&RS=DN/20060031195] US Patent: 20060031195</ref>.

Statistical Compound Term Processing is more adaptive than the "phrase based indexing and retrieval" detailed by Anna Lynn Patterson in her patent applications. The "phrase based indexing" is targeted at searching the World Wide Web where Google can utilise theiran extensive statistical knowledge of common searches can be used to identify candidate phrases. Statistical Compound Term Processing is more suited to [[Enterprise Search]] applications where such a priori knowledge is not available.

Statistical Compound Term Processing is also more adaptive than the linguistic approach taken by the CLAMOUR project which considers the syntactic properties of the terms (part of speech, gender, number) and their combination. CLAMOUR is highly language dependent, whereas the statistical approach is language independent.