Automatic Title Generation in Scientific Articles for Authorship Assistance: A Summarization Approach

Abstract

This paper presents a study on automatic title generation for scientific articles considering sentence information types known as rhetorical categories. A title can be seen as a high-compression summary of a document. A rhetorical category is an information type conveyed by the author of a text for each textual unit, for example: background, method, or result of the research. The experiment in this study focused on extracting the research purpose and research method information for inclusion in a computer-generated title. Sentences are classified into rhetorical categories, after which these sentences are filtered using three methods. Three title candidates whose contents reflect the filtered sentences are then generated using a template-based or an adaptive K-nearest neighbor approach. The experiment was conducted using two different dataset domains: computational linguistics and chemistry. Our study obtained a 0.109-0.255 F1measure score on average for computer-generated titles compared to original titles. In a human evaluation the automatically generated titles were deemed 'relatively acceptable' in the computational linguistics domain and 'not acceptable' in the chemistry domain. It can be concluded that rhetorical categories have unexplored potential to improve the performance of summarization tasks in general.

Sequential Pattern Mining and Deep Learning to Enhance Readability of Indonesian Text Summarization

“…Vector Space Model (VSM) [115], Deep Learning [116], Latent Dirichlet Allocation (LDA) and Genetic Algorithm (GA) [117], Adaptive K-Nearest Neighbor (AKNN) [118], TF-IDF and Phrase Reinforcement algorithm [119], TF-IDF and Machine Learning [120], Naive Bayes Classifier [121], Frequent term based and Sentence weighting [81].…”

“…Term Frequency-Inverse Document Frequency (TF-IDF) is widely used in automatic text summarization [120]; sentence scoring and clustering for text summarization evaluation which proves that clustering is not significantly affect summarization result because the most important sentences are at the beginning of the paragraph (with English document) [111]; analyzing the opinion of consumer from online hotel review with text summarization approach [114]; generating scientific journal article using text summarization approach [118]; deep learning method for text summarization such as deep Auto Encoder and Ensamble Noisy Auto-Encoder [96], Recurrent Neural Networks [110], and neuro-fuzzy approach [29].…”

Semi-automatic System for Title Construction

“…Various studies have been performed on automatic title generation from both spoken [5,11,12] and written text [13,14,22,25,27]. These works aim at converting the document into a 'title representation' by using either statistical, probabilistic, or machine learning methods.…”

Domain Controlled Title Generation with Human Evaluation

“…The title of the research papers becomes really significant due to the availability of scientific papers in abundance. Researchers can automatically determine the importance of a paper by its title rather than reading the whole document [1,2,3]. The accuracy of the title influences the number of potential readers and therefore the number of citations [1,4].…”

“…Researchers can automatically determine the importance of a paper by its title rather than reading the whole document [1,2,3]. The accuracy of the title influences the number of potential readers and therefore the number of citations [1,4]. That's why it's important for the researchers to produce a good title, however, people spend very little time on it [3].…”