Generating key questions from academic texts
Koichi Sanoki · 2017
The search for the efficiency of the use of time dedicated to an activity of reading academic texts is central to the research that is in progress. A prominent contribution of the research is related to an algorithm that generates questionnaires from those texts. This article presents and discusses the main strategy that subsidizes the concerns involved in the development of such algorithm. The strategy will have several stages and the first step of the algorithm will be to interact with the researcher, who should inform the text and the purpose of his research through Keywords. The second step of the algorithm will be to insert this text into a computational medium (if it has not yet been inserted) and from the items of the bibliographic reference it will look for the referenced and related articles in available journals or repositories, in order to create a base of Data that will be called a Bibliographic Corpus which, in turn, will support it. In the creation of Corpus will be generated the morphological and semantic meanings of all words in the text and referenced articles. The next step of the algorithm will be to create relationships between the types of words in the text and the content of the Corpus, taking into account the characteristics of each word. In the stage of semantic relationships of the prayers of the text with the Corpus it will be necessary to separate text and Corpus in sentences, sentences and phrases. From the morphological and semantic meanings of the words of the text will be established the relationships with the adverbs of mode and demonstrative pronouns, which will allow the creation of questions. Finally the last step of the algorithm will be to reconstruct prayers or sentences, adding What, When, Why, Where, Who, How, and How much. The result of applying these steps will be an algorithm for generating questionnaires, in the form of an inquiry, about a particular academic text, and a good reason for holding reading for clarification, or not.