R&D

Recobell zeroes in on solving difficult business problems with the powerful tool of data. Recobell always takes on the challenge of investing in new technology research.

About Our Technology Research

Main Areas of Research
  • Big data
  • Machine
    learning
  • Deep learning
  • AI
  • Text mining
  • Image recognition
Cases of Research Application
  • Real-time personalized recommendation logic optimization
  • IoT platform research
  • Research & Development on ad exposure optimization
  • Image recognition research
  • Research & Development on AI-based marketing operation automization
  • Research & Development on AI Chatbot
  • Research on deriving user interests based on behavior history

Areas of Research

01 Research on AI-based marketing operation automization

Marketers are able to shift their focus from tedious repetitive tasks to more creative thinking and planning.

  • Data-based optimization for experience-driven decision making points
  • Automization for manpower-driven operation management tasks
02 Research & Development on ad exposure optimization

The key to increasing value for user-media-advertiser at the same is in DATA.

  • Real-time prediction and exposure of ad sources that users have interest in
  • Conduct research on data-based prediction logic for major indexes such as click rate and conversion rate
  • Conduct research on exposure logic that will optimize value for all players within the ad ecosystem (user-media-advertiser)
03 Research & Development on AI Chatbot

Increase CS quality and improve CS representative’s work environment through the use of a chatbot that grasps user’s needs and responds appropriately.

  • Conduct research on the analysis of user intentions processed in natural language, the search for data per user intention, and sentence production
04 Research & Development on IoT Platform

Analyze users’ home appliance usage pattern and predict status (such as breakdown) of the home appliances through Big Data platform of IoT home appliances

  • Develop technology that gathers and processes IoT home appliance data
  • Develop usage pattern analysis logic and home appliance status prediction logic
05 Research on deriving user interests based on behavior history

Automatically derive users’ interests through the use of their behavior history and content analysis

  • Conduct research on analysis technology of behavior history logs and texts
  • Conduct related research on interest-based ad exposure optimization
06 Image recognition research

Automatically search and recommend products that users are looking for through image analysis

  • Conduct research on image recognition technology and product similarity valuation technology
SEARCH BY METRIC LEARNING

Learn the metric of classifying similar products in different images and separating products that are completely different by studying various images for each product

Identical products      Different products
CASES OF METRIC LEARNING

Sample case on product matching in an Online Products Dataset: Pictures of the same product that were taken in different angles were matched

(Source : Deep Metric Learning via Lifted Structured Feature Embedding, Hyun Oh Song,  Yu Xiang, Stefanie Jegelka, Silvio Savarese)