A framework for mining depressive symptoms and their relations from consultation records is proposed. The records contain many kinds of depressive symptoms, such as depressed mood, suicide ideas, anxiety, and sleep disturbances. The depressive symptoms are embedded in a single sentence or a discourse segment. The mining task is to identify the discourse segments and their semantic relations. Mining depressive symptoms requires domain knowledge. A framework that integrates semantic-dependency, lexical-cohesion, and domain knowledge sources to mine depressive symptoms and their relations. A semantic dependency model captures the semantics dependency between each word token and its head in a sentence.
All Science Journal Classification (ASJC) codes
- Computer Networks and Communications
- Artificial Intelligence