Conference proceeding
Using a keyword extraction pipeline to understand concepts in future work sections of research papers
17TH INTERNATIONAL CONFERENCE ON SCIENTOMETRICS & INFORMETRICS (ISSI2019), VOL I, pp.87-98
01 Jan 2019
Abstract
This paper presents a methodological framework, based on natural language processing (NLP) techniques, for identifying future work sentences in full-text scientific papers and extracting keywords from these sentences. We conduct a baseline test to evaluate the method's effectiveness and use full-text papers from Science Advances for a proof of concept. Our results suggest that there are significant domain differences in the extent to which keywords in the future work section match those in the title and abstract texts. This is, to our knowledge, the first empirical examination of future work statements in scientific articles. Our framework could contribute greatly to quantitative and predictive studies of science by introducing a new, more future-oriented data source, with potentially significant implications in both theory and practice. Moreover, our proof-of-concept study offers the first piece of evidence about the future work statement as a subgenre of scholarly writing. This evidence may inspire future scientometric studies, leading to a better understanding of the conceptual connections among papers published at different times.
Metrics
18 Record Views
Details
- Title
- Using a keyword extraction pipeline to understand concepts in future work sections of research papers
- Creators
- Kai Li - Drexel Univ, Philadelphia, PA 19104 USAErjia Yan - Drexel Univ, Philadelphia, PA 19104 USA
- Contributors
- G Catalano (Editor)C Daraio (Editor)M Gregori (Editor)H F Moed (Editor)G Ruocco (Editor)
- Publication Details
- 17TH INTERNATIONAL CONFERENCE ON SCIENTOMETRICS & INFORMETRICS (ISSI2019), VOL I, pp.87-98
- Conference
- 17TH INTERNATIONAL CONFERENCE ON SCIENTOMETRICS & INFORMETRICS (ISSI2019), VOL II, 17th
- Series
- Proceedings of the International Conference on Scientometrics and Informetrics
- Publisher
- Int Soc Scientometrics & Informetrics-Issi
- Number of pages
- 12
- Grant note
- RE-07-15-0060-15 / Institute of Museum and Library Services
- Resource Type
- Conference proceeding
- Language
- English
- Academic Unit
- Information Science (Informatics)
- Identifiers
- 991019170385304721
InCites Highlights
These are selected metrics from InCites Benchmarking & Analytics tool, related to this output
- Web of Science research areas
- Computer Science, Interdisciplinary Applications
- Information Science & Library Science