Short Text Similarity

Similarity between different short texts can be computed and the similarity output is a real value within 0 to 1. The higher the value, the better the similarity. This similarity value can be directly used for ranking results as well as more complex systems as a one-dimension basic feature

Free Trial

Benefits

High Accuracy
It provides traditional computing method based on Short Text Similarity an also the similarity computing method based on deep learning
Massive Training Data
Algorithm development and training are made with large-scale mass data, so as to realize deep Chinese semantics comprehension system
High Flexibility
Applications in specific fields and specific scenarios can be rapidly developed and realized according to demands

Features

Text Similarity Depth Model

Multi-model Fusion

Based on multiple computing models, such as the traditional one and the deep learning one, as well as mass training data, similarity between different short texts is computed and the similarity output is a real value within 0 to 1. The higher the value, the better the similarity

Application Scenarios

Wide Scenario Application

It is applicable for Text Categorization, assisting pre-categorizing search; it can recommend products with similar titles according to customer’s browsing records; it can categorize answers and questions, etc.

Scenarios

Vertical Categorization

With Vertical Categorization, similar texts can be retrieved according to texts and pre-categorizing search can be assisted

Product Recommendation

With Product Recommendation, products with similar titles can be detected according to product titles browsed by the customer and recommended to users

Question and Answer Categorization

With Question and Answer Classification, similar questions and answers can be searched for users