A Semantic Frame-Based Similarity Metric for Characterizing Technological Capabilities
- First Online:
In this work we are motivated by the problem of representing technological capabilities that are present in text. We propose to use frames to capture the semantics around technologies and describe a new method, called FrameSim, that serves as a means of determining the similarity between these capabilities. We intentionally focus on a corpus built from informal media (e.g., news articles), which provides greater variability and an increased amount of suppositions about technologies’ uses, deriving value from ‘passive crowdsourcing’. Our evaluation shows that this semantic frame-based similarity metric preserves technology topic coherence, and we discuss how this method shows promise for improving conceptual search in scientific and technical writing.