Latent Semantic Indexing (LSI) uses a patented concept of retrieval which overcomes most of the problems currently associated with the more popular word based retrieval systems. It is an automatic software application that can return 30% more effective relevancy to text based databases than the current standard search word matching techniques used by search engines.
LSI is particularly beneficial when large amounts of data are to be processed and where information needs to be retrieved in multiple languages and translation needed. Current retrieval match words contained within text or documents in a database. while this system is popular they are far from perfect, and if you have ever used a search engine search box you will know what I mean.
The main problem lies in the lack of precision, and on average 50% of the information that is returned will be irrelevant to the end user. there is also the question of retrieval failure whereby only 20% of the relevant information available is returned. In other words, do you know what you are missing?
The main reason for this missing information lies in the fact that there are many ways to describe the same idea or concept. On non-LSI models, if a writer such as I use one word to describe a particular thing, and a searcher such as yourselves use another then you are quite likely to not get all the relevant information available.
An example of this would be if i were to use the word ‘laptop’, yet the searcher looks for ‘notebook’ or ‘portable computer’ it will be more likely to be missing from the search results for that elusive ‘laptop’ keyword. Authors and searchers alike find it increasingly difficult to anticipate the many ways in which a concept can be described.
LSI on the other hand enables the user to find relevant information even if the search query shares no words in common. LSI uses a very powerful and fully automatic way of statistically associating words of description to one another. LSI will learn that “laptop” and “portable” occur in many of the same contexts, and that queries about one should probably retrieve documents about the other.
LSI is 30% more effective than the current word matching methods used in retrieving relevant information to the user. (e.g., Deerwester et al., 1990; Dumais, 1995) LSI can handle many differing types of text based applications such as multimedia descriptions, marketing brochures, email messages or WWW urls all with equal ease. In addition LSI is unique in it’s application to improving information whenever the retrieval rate necessary is high. e.g. multimedia information, advertisements etc.
Because LSI does not rely on direct word matching it finds more relevant information at a breeze.even in differing languages without any necessary changes to existing algorithms. No wonder the likes of Google are now heading in the LSI direction, it’s search results can only improve on the relevancy to it’s clients. i.e you and me.