Search

Building dialogue systems that can converse naturally with humans is a challenging yet intriguing problem of artificial intelligence. Dialogue system is categorized into task-oriented system and open-domain system. In open-domain dialogue system, the system is expected to respond to human utterances in an interesting and engaging way. With the development of deep learning and the availability of large corpora, there is a trend towards developing generative dialogue systems. However, current generative dialogue system is learned directly from the dialogue corpus and ignores the common sense knowledge, which makes it difficult to respond substantively. Hence, common sense knowledge has to be effectively integrated in the generative dialogue system. Knowledge graph is a structured data to represent the common sense knowledge. In this thesis, we investigate the impact of connecting knowledge graph with the generative dialogue system. In single-turn setting, we propose generative dialogue system (GenDS) which can generate the reply with multiple knowledge triples. Besides, GenDS does not rely on the representations of entities, thus can handle out-of-vocabulary entities. In the multi-turn scenario, we propose a multi-turn GenDS which jointly takes into account message history and related common sense for generating a coherent response. We collect a human-to-human conversation data (ConversMusic) with knowledge annotations. The proposed systems are evaluated on CoversMusic and a public question answering dataset. Our proposed systems outperform baseline methods significantly in terms of the BLEU, entity accuracy, entity recall and human evaluation.

The prediction of academic performance is one of the most important tasks in educational data mining, and has been widely studied in MOOCs and intelligent tutoring systems. Academic performance could be affected with factors like personality, skills, social environment, the use of library books and so on. However, it is still less investigated that how could the use of library books affect academic performance of college students and even leverage book-loan history for predicting academic performance. To this end, we propose a supervised contentaware matrix factorization for mutual reinforcement of academic performance prediction and library book recommendation. This model not only addresses the sparsity challenge by explainable dimension reduction techniques, but also promotes library book recommendation by recommending "right" books for students based on their performance levels and book meta information. Finally, we evaluate the proposed model on three years of the book-loan history and cumulative grade point average of 13,047 undergraduate students in one university. The results show that the proposed model outperforms the competing baselines on both tasks, and that academic performance is not only predictable from the book-loan history but also improves the recommendation of library books for students. © 2016 IEEE.

In knowledge grounded conversation, domain knowledge plays an important role in a special domain such as Music. The response of knowledge grounded conversation might contain multiple answer entities or no entity at all. Although existing generative question answering (QA) systems can be applied to knowledge grounded conversation, they either have at most one entity in a response or cannot deal with out-of-vocabulary entities. We propose a fully data-driven generative dialogue system GenDS that is capable of generating responses based on input message and related knowledge base (KB). To generate arbitrary number of answer entities even when these entities never appear in the training set, we design a dynamic knowledge enquirer which selects different answer entities at different positions in a single response, according to different local context. It does not rely on the representations of entities, enabling our model deal with out-of-vocabulary entities. We collect a human-human conversation data (ConversMusic) with knowledge annotations. The proposed method is evaluated on CoversMusic and a public question answering dataset. Our proposed GenDS system outperforms baseline methods significantly in terms of the BLEU, entity accuracy, entity recall and human evaluation. Moreover,the experiments also demonstrate that GenDS works better even on small datasets.