Aiming at the identification of Chinese fine-grained implicit discourse relation and taking the directionality characteristic in account, the authors propose a feature learning algorithm based on the distant supervision to label explicit discourse data automatically. The relative position information between conjunction and words are applied to train the intensive word representation. Then the rhetorical function of words and the directionality of relations are encoded into the representation of intensive words, which is applied to the relation classification of fine-grained implicit discourses. From the experimental studies of the proposed approach, the classification accuracy reaches 49.79%, which are better than those approaches neglecting the directionality of discourse relations.
Based on the study of retrieving plane geometric figures (PGFs) in the area of computer aided instruction, a feasible solution for PGF retrieval is proposed. The authors focus on several challenging tasks such as sketch beautification, geometric primitive detection, salience analysis of the overlapped primitives, structural relationship description between two geometric primitives, and figure similarity computing. Several algorithms are presented especially on layout description and complex shape matching. The PGFs are applied directly to content retrieval and compensate for the weaknesses in describing the query intentions using keyword-based search. Experimental results demonstrate the feasibility and significant performance of the proposed retrieval algorithm.