A computer program product comprising a non-transitory computer-readable medium including programmed instructions, the instructions causing a computer to execute:
- extracting, from each of a plurality of documents, a plurality of causal relationships, each of the causal relationships including a set of a first element and a second element having a relationship therebetween;
- registering the extracted causal relationships in causal relationship management information;
- learning a feature amount of each of a plurality of words included in each of the plurality of documents;
- registering the feature amounts in feature management information;
- acquiring a causal relationship included in a target document from the causal relationship management information;
- acquiring, with respect to the causal relationship included in the target document, a similar expression having a similar feature amount to at least one of a feature amount of the first element and a feature amount of the second element included in the causal relationship based on the feature management information;
- determining a priority of each of words registered in the feature management information;
- acquiring, among the words registered in the feature management information, a group of words having a higher priority than a predetermined first priority from words having a similar feature amount to the feature amount of the first element or the feature amount of the second element that constitutes the causal relationship in the target document as the similar expression;
- and
- acquiring, with respect to the causal relationship included in the target document, a generalized expression obtained by generalizing an expression of the at least one of the first element and the second element included in the causal relationship,
- wherein the generalized expression is not limited to a specific expression within a category.