Because we work to help Recruit improve its 200 customer facing tools, our research finds the gaps and provides advancement tools for machine learning, natural language processing, and database management. We aim to use our findings to produce solutions applicable to real world problems. Machine learning has been a core component of our research as we embark to developed projects to assist researchers with database management, data analysis, natural language processing and in finding technologies to facilitate related tasks.
As part of our goal to innovate for the users of Recruit Holdings’ many products, our researchers set out to create a subjective database that allows for a user to search for hotel, restaurants and attractions with subjective queries. From this work we developed OpineDB and it’s front end Voyageur. OpineDB serves results based on aggregated reviews, thus Voyageur allows users to search subjectively and save time digging through thousands of reviews.
With the goal of building technology that can understand how people express their happy moments in text, we crowd-sourced HappyDB, a corpus of 100,000 happy moments that we make publicly available. HappyDB helped us outline several important natural language processing problems that can be studied with the help of the corpus. We also apply several state-of-the-art analysis techniques to analyze HappyDB. Our results demonstrate the need for deeper NLP techniques to be developed which makes HappyDB an exciting resource for follow-on research.
MegaMiner is a toolbox that we’re currently building to extract insights from user-generated text such as reviews. Reviews are a ubiquitous ingredient of e-commerce applications and developing models and algorithms to leverage the information in review text is a core research activity at Megagon Labs. MegaMiner will include a suite of tools: intelligent text labeling for supporting model development, interactive exploratory text analysis, opinion extraction, text summarization and aggregation.