Semi-automatic System for Title Construction

In this paper, we introduce a two-phase, semi-automatic system for title construction. Our work is based on the hypothesis that keywords are good candidates for title construction. The two phases of the system consist of extracting keywords from the document and constructing a title using these keywords. The proposed system does not generate the title, instead it aids the author in creatively constructing the title by suggesting impactful words. The system uses a pre-trained supervised keyword extraction model to extract important words from the text. Our KE approach gains from the advantages of graph-based keyword extraction techniques. We empirically establish the effectiveness of the proposed method, and show that it can be applied to any texts across domain and corpora. The keywords thus extracted are suggested to the author as potential candidates for inclusion in the title. The author can use creative transformations of the suggested words to construct an appropriate title for the manuscript. We evaluate the proposed system by computing the overlap between the list of title-words and the extracted keywords from the documents, and observe a macro-averaged precision of 82%.