Latent Semantic Indexing
In regular keyword searching process the Search Engines display a collection of results where the documents contain the word placed for a query. The traditional search process creates a result processed by looking through each document for certain keyword or keyword phrase in the query box. The documents do not have any kind of interdependence between them and are evaluated basing on their contents in Search Engine Result Pages.
Latent Semantic Indexing process gives a new direction to the process of indexing document. Latent Semantics Indexing is an exclusive information retrieval method in search engines. It is an approach to automated document indexing deviating from the word-based indexing system.
Using statistical algorithm, LSI can retrieve relevant documents though they don’t contain any word of the query. LSI considers the documents that have words semantically close with the keywords or keyword phrases searched as queries. It allows the search engines to know what a page is all about. It places additional weight on related words in the content and hence, lowers the value of the page that only matches the exact terms.
Although LSI doesn’t understand the meaning of the words, it uses a fully automatic statistical method to reveal the connection among the terms in a large collection of texts. The method is called single value decomposition. The advantage of LSI is that it is totally a mathematical approach. This makes LSI a powerful and generic tool to index any interconnected collection of documents in any language. It is a more effective method than the popular word-matching method to provide relevant information to the users. It can be used with regular keyword search. Apart from being automatic and widely applicable, it can handle trouble reports, multimedia descriptions, email massages, marketing brochures.
Applications of LSI:
Overcoming the drawbacks of traditional Search Engine Indexing techniques, the process of LSI provides significant results. LSI has given a different dimension to Search Engine Result Pages. The function of LSI can be discussed as:
Comparison of Data
LSI compares documents and classifies the data into different groups.
Relevant Data
Unlike traditional search process, LSI not only searches for the particular word, but also for the documents related to that word. Therefore, it is able to display a list of documents that have a great deal of relevancy with the searched keywords. LSI can work with a various set of keywords and give better result as the more it knows about a document the better it is in finding similar documents. It is a more interactive search with the users producing useful result.
Textual Consistency
LSI can calculate the degree of logical relationship between its ingredients of texts by looking their semantic relationships. This kind of logical relationship correlates readability and comprehension and this makes LSI a useful tool in writing instruction.
Filtered Information
With the LSI the information becomes more filtered as Search Engine Result Page displays collection of documents that are sharing logical relation with the search query.
Multiple Languages
LSI is applicable when there is a need of retrieval of information in multiple languages without requiring translation of queries or document.