支撑软件安全评估的自然语言数据分散在各个异构的交流平台和交流软件中，这

The natural language data supporting software safety assessment are scattered in various heterogeneous communication platforms and communication software, which leads to the fact that we cannot directly obtain these natural language data in a simple way. For example, the security feature description of some systems exists in the user’s formal software defect report, while the description information of other software may be implicit in the user’s technical exchange forum. This requires our security assessment methods to be able to break through these intelligently. Heterogeneous platforms limit the collection of natural language data.

The natural language data supporting software security assessment are scattered in various heterogeneous communication platforms and communication software, which makes it impossible for us to directly obtain these natural language data in a simple way. For example, the security characteristics of some systems exist in the user's formal software defect report, The description information of other software may be hidden in the user's technical communication forum, which requires that our security evaluation method must be able to intelligently break through the restrictions of these heterogeneous platforms to collect natural language data<br>

The natural language data supporting software security assessment are scattered in various heterogeneous communication platforms and communication software, which leads us to be unable to obtain these natural language data directly in a simple way. For example, the description of security characteristics of some systems exists in the formal software defect reports of users, while the description information of other software may be hidden in the technical communication forums of users, which requires that our security assessment method must be able to intelligently break through the limitations of these heterogeneous platforms to collect natural language data.