Several approaches to extract keywords from a document submitted by students were done. Two types of documents were used. Summaries of research papers were analyzed as the fixed structure type. Another type was the non-fixed structure type. Non-significant words that had little meaning were discarded after morphological analysis. Remained words were used as extracted terms. Frequencies of terms in the document, associations of terms in the sentence, and weight-degrees of terms in the documents were examined. Chart-ization was tried using these values. Documents of the non-fixed structure type were chart-ized with weight-degrees of terms. All but a few of significant terms were extracted from documents of the fixed structure type, and we could not chart-ize these documents.