Majority Internet search engines up to today do keywords searching. User picks up few keywords for what they want to search for. Search engine finds the web pages stored in its database that contain these keywords and sorts the result in certain way and then return them back to user. This approach finds all the web pages has the keywords, however, not necessarily the web pages that user want to read. In most of cases, relevant web pages are a subset of the search results.

SRIR technology enables search not only by key words, but the syntax relationship between key words as well. SRIR allows question/answer (Q/A) type of search. This is achieved by first analyze the semantics of search questions and then create a answer semantic pattern that in turn is used to search in Infomation Database (InfoDB) for relevant answers. Web pages’ content is first go through a syntax analyzer. Text semantic is recognized in a language called component and pattern description language (CPDL). When user submit a search request, it may include predefined keywords relationship and semantic descriptor that will be used to identify the web pages has the same pattern. SRIR technology can also be used to realize information agent (IA). User tells the IA what kind of information are interested. IA searchs internet periodically and find relevent web pages, create and return digest for the searched pages.

SRIR Overview

SRIR is consisted of three major parts. Component and pattern description language (CPDL) is used to describe text components, their properties as well as syntax and semantic patterns. Pattern Procesing Engine (PPE) parses searched web pages and create pattern model using CPDL. Pattern model describe in CPDL is stored in InfoDB, which is database used by SRIR to provide requested information.