Technology relationship analysis using problem and solution similarities

In this research, we divided the patent similarity into the problem similarity and the solution similarity to define specific technology relationships. To study the possibility, the significance, and the application areas, we collected 2,050 patents and 2,100,225 patent pairs as a sample data. After studying the patents, we developed methods for the text extraction and the similarity calculation. Utilizing calculated similarities, we conducted the correlation analysis and the paired-samples t-test to prove the problem similarity and the solution similarity are distinctive. Then we classified patent pairs into 4 groups through the clustering algorithm. Finally, we studied characteristics of the groups and discovered application areas.