The Implementation And Research Of The Probabilistic Latent Semantic Analysis Model In The Search Engine's Business Text Classification System

Subject to search engine profitability, the correlation between the user's intention and the advertisement ejected by the search engine is very important. Application of the traditional text classification method in commercial text classification system has provided us some help but very limited due to the semantic abstractness, polysemy and synonymy. How to define and calculate the semantic of words, phrases and sentences in its context remains a big commercial demand of search engine but has not been satisfied. Aiming at the demand, the author of the paper, under the guidance of principle of software engineering, using Probabilistic Latent Semantic Analysis (PLSA) method, designed a probabilistic latent semantic computation module in commercial text classification system in search engine. At the end, according to business requirements of commercial search engine, the author tested the module under relevant testing protocols. The test result approves it is an effective and practical module.