Intelligence Information Retrieval in Corporate Information Systems

Local searching systems development is not a trivial task. The amount of data is too large for an exhaustive search, but not enough for the Bigdata algorithms. Some local searching systems built using existing searching engines could have low relevance for a specific user. Therefore, we need to integrate special algorithms for increasing the quality of searching. In this paper, we consider the ways of applying the machine-learning methods such as topic modeling and neural networks and demonstrate how we can increase search quality, based on the actual statistics collected. In addition, we demonstrate the evaluation criteria, based on the collected statistics, which we developed because of the unavailability of standard ways of evaluating the quality of the search engine.