Diyi Yang

Hierarchical Attention Networks for Document Classification

We propose a hierarchical attention network for document classification. Our model has two distinctive characteristics: (i) it has a hierarchical structure that mirrors the hierarchical structure of documents; (ii) it has two levels of attention mechanisms applied at the wordand sentence-level, enabling it to attend differentially to more and less important content when constructing the document representation. Experiments conducted on six large scale text classification tasks demonstrate that the proposed architecture outperform previous methods by a substantial margin. Visualization of the attention layers illustrates that the model selects qualitatively informative words and sentences.

Social factors that contribute to attrition in MOOCs

In this paper, we explore student dropout behavior in a Massively Open Online Course (MOOC). We use a survival model to measure the impact of three social factors that make predictions about attrition along the way for students who have participated in the course discussion forum.

Exploring the Effect of Confusion in Discussion Forums of Massive Open Online Courses

Thousands of students enroll in Massive Open Online Courses~(MOOCs) to seek opportunities for learning and self-improvement. However, the learning process often involves struggles with confusion, which may have an adverse effect on the course participation experience, leading to dropout along the way. In this paper, we quantify that effect. We describe a classification model using discussion forum behavior and clickstream data to automatically identify posts that express confusion. We then apply survival analysis to quantify the impact of confusion on student dropout. The results demonstrate that the more confusion students express or are exposed to, the lower the probability of their retention. Receiving support and resolution of confusion helps mitigate this effect. We explore the differential effects of confusion expressed in different contexts and related to different aspects of courses. We conclude with implications for design of interventions towards improving the retention of students in MOOCs.

Incorporating Word Correlation Knowledge into Topic Modeling.

This paper studies how to incorporate the external word correlation knowledge to improve the coherence of topic modeling. Existing topic models assume words are generated independently and lack the mechanism to utilize the rich similarity relationships among words to learn coherent topics. To solve this problem, we build a Markov Random Field (MRF) regularized Latent Dirichlet Allocation (LDA) model, which defines a MRF on the latent topic layer of LDA to encourage words labeled as similar to share the same topic label. Under our model, the topic assignment of each word is not independent, but rather affected by the topic labels of its correlated words. Similar words have better chance to be put into the same topic due to the regularization of MRF, hence the coherence of topics can be boosted. In addition, our model can accommodate the subtlety that whether two words are similar depends on which topic they appear in, which allows word with multiple senses to be put into different topics properly. We derive a variational inference method to infer the posterior probabilities and learn model parameters and present techniques to deal with the hardto-compute partition function in MRF. Experiments on two datasets demonstrate the effectiveness of our model.

Positive Impact of Collaborative Chat Participation in an edX MOOC

A major limitation of the current generation of MOOCs is a lack of opportunity for students to make use of each other as resources. Analyses of attrition and learning in MOOCs both point to the importance of social engagement for motivational support and overcoming difficulties with material and course procedures. In this paper we evaluate an intervention that makes synchronous collaboration opportunities available to students in an edX MOOC. We have implemented a Lobby program that students can access via a live link at any time. Upon entering the Lobby, they are matched with other students that are logged in to it. Once matched, they are provided with a link to a chat room where they can work with their partner students on a synchronous collaboration activity, supported by a conversational computer agent. Results of a survival model in which we control for level of effort suggest that having experienced a collaborative chat is associated with a slow down in the rate of attrition over time by a factor of two. We discuss implications for design, limitations of the current study, and directions for future research.

Predicting advertiser bidding behaviors in sponsored search by rationality modeling

We study how an advertiser changes his/her bid prices in sponsored search, by modeling his/her rationality. Predicting the bid changes of advertisers with respect to their campaign performances is a key capability of search engines, since it can be used to improve the offline evaluation of new advertising technologies and the forecast of future revenue of the search engine. Previous work on advertiser behavior modeling heavily relies on the assumption of perfect advertiser rationality; however, in most cases, this assumption does not hold in practice. Advertisers may be unwilling, incapable, and/or constrained to achieve their best response. In this paper, we explicitly model these limitations in the rationality of advertisers, and build a probabilistic advertiser behavior model from the perspective of a search engine. We then use the expected payoff to define the objective function for an advertiser to optimize given his/her limited rationality. By solving the optimization problem with Monte Carlo, we get a prediction of mixed bid strategy for each advertiser in the next period of time. We examine the effectiveness of our model both directly using real historical bids and indirectly using revenue prediction and click number prediction. Our experimental results based on the sponsored search logs from a commercial search engine show that the proposed model can provide a more accurate prediction of advertiser bid behaviors than several baseline methods.

Humor Recognition and Humor Anchor Extraction

Humor is an essential component in personal communication. How to create computational models to discover the structures behind humor, recognize humor and even extract humor anchors remains a challenge. In this work, we first identify several semantic structures behind humor and design sets of features for each structure, and next employ a computational approach to recognize humor. Furthermore, we develop a simple and effective method to extract anchors that enable humor in a sentence. Experiments conducted on two datasets demonstrate that our humor recognizer is effective in automatically distinguishing between humorous and non-humorous texts and our extracted humor anchors correlate quite well with human annotations.

That's So Annoying!!!: A Lexical and Frame-Semantic Embedding Based Data Augmentation Approach to Automatic Categorization of Annoying Behaviors using #petpeeve Tweets

We propose a novel data augmentation approach to enhance computational behavioral analysis using social media text. In particular, we collect a Twitter corpus of the descriptions of annoying behaviors using the #petpeeve hashtags. In the qualitative analysis, we study the language use in these tweets, with a special focus on the fine-grained categories and the geographic variation of the language. In quantitative analysis, we show that lexical and syntactic features are useful for automatic categorization of annoying behaviors, and frame-semantic features further boost the performance; that leveraging large lexical embeddings to create additional training instances significantly improves the lexical model; and incorporating frame-semantic embedding achieves the best overall performance.

Towards Identifying the Resolvability of Threads in MOOCs

One important function of the discussion forums of Massive Open Online Courses (MOOCs) is for students to post problems they are unable to resolve and receive help from their peers and instructors. There are a large proportion of threads that are not resolved to the satisfaction of the students for various reasons. In this paper, we attack this problem by firstly constructing a conceptual model validated using a Structural Equation Modeling technique, which enables us to understand the factors that influence whether a problem thread is satisfactorily resolved. We then demonstrate the robustness of these findings using a predictive model that illustrates how accurately those factors can be used to predict whether a thread is resolved or unresolved. Experiments conducted on one MOOC show that thread resolveability connects closely to our proposed five dimensions and that the predictive ensemble model gives better performance over several baselines.

Informative household recommendation with feature-based matrix factorization

In this paper, we describe our solutions to the first track of CAMRa2011 challenge. The goal of this track is to generate a movie ranking list for each household. To achieve this goal, we propose to use the ranking oriented matrix factorization and the matrix factorization with negative examples sampling. We also adopt feature-based matrix factorization framework to incorporate various contextual information to our model, including user-household relations, item neighborhood, user implicit feedback, etc. Finally, we elaborate two kinds of methods to recommend movies for each household based on our models. Experimental results show that our proposed approaches achieve significant improvement over baseline methods.