In our presentation we show a method, which makes possible to measure precision of keyword searches executed in full text databases. This method analyses how much information on average is expressed by the context of keywords in connection with a specific keyword in the database. Since the average information content depends on other elements of the database, thus we can consider this method objective. Using this method we can create user types, which categorize people who carry out various searches. We place those individuals in the first category who search for novelties, so they want to find texts with high average information-content. Those persons belong to the second category who search widespread relationships of meaning, as they wish to obtain texts with a low average information-content. In our test we determine the technical relevance of search results, but we take into consideration the user needs through the created user types. Previously we suppose that the average informationcontent of a textual document reflects its precision. In our analysis we examine this hypothesis in more details. Another interesting question, which emerges during our analysis is how we can use notions of technical relevance and technical precision at keyword searches in a full text database. Finding appropriate answer to this question makes possible that objective and really mathematical methods would appear in the relevance measurement of keyword searches in order to check rather subjective methods.