This work is dedicated to an actual problem of efficient information search in the Internet. The work is based on the algorithms of sentences comparison taking into account the schemes of syntactic analysis generated by Link Grammar Parser software. The main idea is that syntactic diagrams give us a primitive structure of a text, which allows us to select phrases in a text which have a syntactic structure similar to that given in a request. According to these ideas, the Information Retrieval System (IRS) iNetSearch was developed. Our study showed that it is often sufficient to remain on the syntactic level and obtain rather good search results. The final part of the article represents the results of testing for the methods implemented within iNetSearch.