Cheng, W.-N., & Khoo, C.S.G. (2021). Information structures in sociology research papers: Modeling cause–effect and comparison relations in research objective and result statements. Journal of the Association for Information Science & Technology, 72(11), 1367-1385.

Research Relation Frame

Abstract. When writing a research paper, the author has to select information to include in the paper to support various arguments. The information has to be organized and synthesized into a coherent whole through relationships and information structures. There is hardly any research on the information structure of research papers, and how information structure supports rhetorical and argument structures. Thus, this study is focused on information organization in the Abstract and Introduction sections of sociology research papers, analyzing the information structure of research objective, question, hypothesis, and result statements. The study is limited to research papers reporting research that investigated cause–effect relations between two concepts. Two semantic frames were developed to specify the types of information associated with cause–effect and comparison relations, and used as coding schemes to annotate the text for different information types. Six link patterns between the two frames were identified—showing how comparisons are used to support the claim that the cause-effect relation is valid. This study demonstrated how semantic frames can be incorporated in discourse analysis to identify deep structures underlying the argument structure. The results carry implications for the knowledge representation of academic research in knowledge graphs, for semantic relation extraction, and teaching of academic writing.

Comparison Frame

The six link patterns are as follows:

  • Link pattern 1: Comparison of Concept1 (cause concept) subclasses/attributes/aspects
    Comparison of two subclasses, attributes or aspects of concept1 (the cause concept) based on their scores on some criterion attribute related to concept2 (the effect concept) is commonly used to establish a Research relation between concept1 and concept2 in a research result statement. The Comparison result can be an attribute value of concept2, or the polarity and/or size of the Research-relation.
  • Link pattern 2: Comparison of Concept2 (effect concept) subclasses, attributes, or aspects
    In research objective and result statements, there is occasionally a comparison of the effect concept’s attributes/aspects/subclasses. The comparison result is that the two effect attributes have different cause concepts.
  • Link pattern 3: Comparing two subclasses of moderator and mediator variables
    Research results are sometimes generated by comparing two subclasses of a moderator or, more rarely, a mediator variable. The measure and comparison result are often concept2 (effect concept) of the Research-relation.
  • Link pattern 4: Comparing different underlying explanations
    In a research result statement, a comparison may link two Research-relation frames with related cause concepts (which may be subclasses of a broader concept), but is focused on highlighting different underlying explanations (the comparison result).
  • Link pattern 5: Comparing results to expectation, hypothesis, or theory
    The research result statement may compare the study result regarding a Research-relation with commonsense expectation, or the result predicted by a hypothesis statement or derived from a theory.
  • Link pattern 6: Comparing results to those of previous study
    The research result statement may contrast the study results regarding a Research-relation with the results of one or more previous studies