Optimization of similarity algorithm for short texts

As a technical member

• Project Description: Provide a scalable solution for short text similarity calculation.
• Responsible part: Based on Baidu know platform Q&A dataset, optimize the R-gram method based on long text, i.e., randomize and extract part of sliding window for matching, based on which, integrate the corresponding knowledge base by domain to give window weight, which improves the matching efficiency by 90% compared with the traditional method of deep learning.