TOPIX: AI-BASED SOLUTION ALLOWING TO SUMMARIZE MASSIVE DATABASES INVOLVING TEXT WRITTEN BY USERS ON PRODUCTS

The AI-based solution produces a simultaneous clustering (co-clustering) of users and products, based on the key emerging topics from the reviews and by the underlying model. The typical use case consists in an e-commerce company interested in understanding the relationship between its users and the sold products thanks to the analysis of user comments. While existing solutions focus on analyzing the fact that users have bought specific products and conversely, other solutions can analyze huge amount of text data to extract useful information, TOPIX is a unique technology that can do both by simultaneously analyzing the links between users and products and the associated texts such as reviews, comments or opinions.

Applications

• E-commerce
• Marketing
• Healthcare
• Science watch
• Social network, e-journalism

Competitive advantages

• Simultaneous and automatic clustering (co-clustering) of rows (ex.: users), columns (ex.: products) based on the key topics (identified from the words analyzed in the texts)
• Produces key topics from the reviews and by the underlying model
• Massive and possibly extremely sparse datasets
• Unsupervised AI
• Image pre-processing routine allowing TOPIX to deal with images (in addition to text)
• Four types of graphs proposed to the user for the outcomes

Intellectual property

Filling of the software

Keywords

Artificial Intelligence - Co-clustering - Latent Topic Block Model (LTBM) - Variational inference algorithm - Textual data analysis - E-commerce