Patent-Based Technology Efficacy Information Extraction in Product Innovation Design

Patent literature is characterized by redundant information, unstructured and different values, and it is difficult to directly obtain the technology realization path composed of technical subject words and efficacy information in patents. In view of the characteristics that the distribution of efficacy information in technical realization paths is more concentrated in patent abstracts, this paper firstly determines to adopt Bi-LSTM-Attention-CRF model for information extraction; then preprocesses the extracted subject words and constructs a feature dictionary to locate the statements with efficacy information; secondly, in order to improve the accuracy of technical subject word extraction, constructs the technical subject word features, and extracts the information based on the BI-LSTM-CRF model. The technical realization path of patents is constructed through the technical subject word-efficacy information, and then technical information such as technical solutions and technical hotspots to meet the user's needs are obtained.