Abstract: Methods, software and devices for responding to a user query are disclosed. A user query received from a user is matched to stored responses to obtain initial matched responses. A confidence metric indicating the likelihood of the initial matched responses satisfying the user query is calculated. Upon determining the confidence metric to be below a specified threshold, the user query is supplemented with contextual data. The user query, supplemented by contextual data, is further matched to stored responses by applying disparate matching techniques to obtain further matched responses, where at least one of the disparate matching techniques differs from techniques used to obtain the initial matched responses. Degree-of-match metrics are calculated taking into account the contextual data, where each degree-of-match metric indicates a quality of match between the user query and a matched response. At least one of the matched responses is presented to the user.

Abstract: A computerized method of analyzing a knowledgebase comprising; assembling a collection of queries made by users to obtain information from the knowledgebase; identifying in each query, sets of collocated words in that query to form a list of collocated word sets in the collection; from the list, identifying and presenting frequently collocated word sets in the collection. Likewise, a histogram of scaled relative difference between the frequency of word sets at first and second time intervales may be presented.

Abstract: Methods, software and devices for responding to a user query are disclosed. A user query received from a user is matched to stored responses to obtain initial matched responses. A confidence metric indicating the likelihood of the initial matched responses satisfying the user query is calculated. Upon determining the confidence metric to be below a specified threshold, the user query is supplemented with contextual data. The user query, supplemented by contextual data, is further matched to stored responses by applying disparate matching techniques to obtain further matched responses, where at least one of the disparate matching techniques differs from techniques used to obtain the initial matched responses. Degree-of-match metrics are calculated taking into account the contextual data, where each degree-of-match metric indicates a quality of match between the user query and a matched response. At least one of the matched responses is presented to the user.

Abstract: Methods, software and devices for indexing responses for later providing to users in response to queries are disclosed. For each stored response, representative queries are stored in association with that response, where each representative query represents a possible query for searching for information addressed by that response. Representative queries are selectively modified by substituting terms by corresponding chosen substitute expressions, where a substitute expression is chosen for a particular term in one of the representative queries based on past substitutions in others of said representative queries. For each response, a Boolean expression is formed from those representative queries associated with that response, as selectively modified, where the Boolean expression is satisfied by each of those representative queries.

Abstract: A computer implemented method of indexing a plurality of responses for later retrieval and presentation to a user in response to queries, includes, for each of the plurality of responses, receiving at least one representative query for that response. The representative query(s) represent text (e.g. natural language) query(s) to be input by an end user searching for information addressed by that response. Each representative query is parsed into terms. The terms are analyzed to determine which of these terms are more likely to uniquely identify queries for the particular response among terms in representative queries for all indexed responses. Boolean expression(s) satisfied by a text query containing one of the terms determined to more likely uniquely identify that response, and another one of the parsed terms are formed.

Abstract: A computer implemented method of indexing a plurality of responses for later retrieval and presentation to a user in response to queries, includes, for each of the plurality of responses, receiving at least one representative query for that response. The representative query(s) represent text (e.g. natural language) query(s) to be input by an end user searching for information addressed by that response. Each representative query is parsed into terms. The terms are analyzed to determine which of these terms are more likely to uniquely identify queries for the particular response among terms in representative queries for all indexed responses. Boolean expression(s) satisfied by a text query containing one of the terms determined to more likely uniquely identify that response, and another one of the parsed terms are formed.