Learning Summarization by Using Similarities

Summarization is an important skill that many of us need. Yet, this skill must be learned. This is why it is taught as a part of composition courses in language programs. The final goal of our research project is to design a system to help students to learn to summarize French texts by using a method called Case-Based Reasoning (CBR). Such a system uses previous, similar situations in order to make a new summary by reusing and adapting, if necessary, the same summarization rules. It then saves the new case—the input text, its summary and the applied summarization rules—which will serve for future, analogous situations. Therefore, each time the system builds a summary, an example of the summarization process is created. One can learn how to make a summary by observing the system's method of summarization, and by getting the system-generated explanations which accompany the process. This paper explores this new research topic and describes the rationale underlying the learning of text summarization by using CBR.