FHKG: A Framework to Harvest Knowledge from Groupware Raw Data for AI
Chukwudi Festus Uwasomba, Yunli Lee, Zaharin Yusoff, Chin Teck Min · 2021
In the era of textual data explosion, including due to a rising remote work culture, a system to harvest on-the-job knowledge of experts from groupware for AI enrichment has become one of the crucial technologies sought after in the field of knowledge technology. Most existing systems for knowledge harvesting are developed based on text corpora from the web, social media, newspapers and textbooks with little or no changeable modules and ontological representations. In this paper, we propose a deeper framework with changeable modules to acquire and represent knowledge from raw data in groupware discussions for AI. Such a framework can be implemented on any platform of choice using existing or newly designed modules that can be continually improved upon with higher sophistication or by added-value extensions. The framework is a formalisation of a semi-automated structure with reusable and incremental modules. The overall architecture of the framework is presented with evaluation results. The paper concludes by highlighting the proposed future developments within the framework.