Options
Dong, Ruihai
Preferred name
Dong, Ruihai
Official Name
Dong, Ruihai
Research Output
Now showing 1 - 10 of 26
- PublicationLeveraging BERT to Improve the FEARS Index for Stock ForecastingFinancial and Economic Attitudes Revealed by Search (FEARS) index reflects the attention and sentiment of public investors and is an important factor for predicting stock price return. In this paper, we take into account the semantics of the FEARS search terms by leveraging the Bidirectional Encoder Representations from Transformers (BERT), and further apply a self-attention deep learning model to our refined FEARS seamlessly for stock return prediction. We demonstrate the practical benefits of our approach by comparing to baseline works.
534 - PublicationMulti-level Attention-Based Neural Networks for Distant Supervised Relation ExtractionWe propose a multi-level attention-based neural network forrelation extraction based on the work of Lin et al. to alleviate the problemof wrong labelling in distant supervision. In this paper, we first adoptgated recurrent units to represent the semantic information. Then, weintroduce a customized multi-level attention mechanism, which is expectedto reduce the weights of noisy words and sentences. Experimentalresults on a real-world dataset show that our model achieves significantimprovement on relation extraction tasks compared to both traditionalfeature-based models and existing neural network-based methods
295 - PublicationFrom More-Like-This to Better-Than-This: Hotel Recommendations from User Generated ReviewsTo help users discover relevant products and items recommender systems must learn about the likes and dislikes of users and the pros and cons of items. In this paper, we present a novel approach to building rich feature-based user profiles and item descriptions by mining user-generated reviews. We show how this information can be integrated into recommender systems to deliver better recommendations and an improved user experience.
349Scopus© Citations 9 - PublicationCombining similarity and sentiment in opinion mining for product recommendation(Springer, 2016-04)
; ; ; ; In the world of recommender systems, so-called content-based methods are an important approach that rely on the availability of detailed product or item descriptions to drive the recommendation process. For example, recommendations can be generated for a target user by selecting unseen products that are similar to the products that the target user has liked or purchased in the past. To do this, content-based methods must be able to compute the similarity between pairs of products (unseen products and liked products, for example) and typically this is achieved by comparing product features or other descriptive elements. The approach works well when product descriptions are readily available and when they are detailed enough to afford an effective similarity comparison. But this is not always the case. Detailed product descriptions may not be available since they can be expensive to create and maintain. In this article we consider another source of product descriptions in the form of the user-generated reviews that frequently accompany products on the web. We ask whether it is possible to mine these reviews, unstructured and noisy as they are, to produce useful product descriptions that can be used in a recommendation system. In particular we describe a novel approach to product recommendation that harnesses not only the features that can be mined from user-generated reviews but also the expressions of sentiment that are associated with these features. We present a recommendation ranking strategy that combines similarity and sentiment to suggest products that are similar but superior to a query product according to the opinion of reviewers, and we demonstrate the practical benefits of this approach across a variety of Amazon product domains.2146Scopus© Citations 53 - PublicationWhy I like it: Multi-task Learning for Recommendation and ExplanationWe describe a novel, multi-task recommendation model, which jointly learns to perform rating prediction and recommendation explanation by combining matrix factorization, for rating prediction, and adversarial sequence to sequence learning for explanation generation. The result is evaluated using real-world datasets to demonstrate improved rating prediction performance, compared to state-of-the-art alternatives, while producing effective, personalized explanations.
1616Scopus© Citations 97 - PublicationUP-TreeRec: Building dynamic user profiles tree for news recommendationOnline News recommendation systemsaim to address the information explosion of news and make personalized recommendation for users. The key problem of personalized news recommendation is to model users’ interests and track their changes. A common way to deal with usermodeling problem is to build user profiles from observed behavior. However, the majority of existing methods make static representation to user profiles and little research has focused on the effective user modeling that could dynamically capture user interests in news topics.To address this problem, in this paper, we propose UP-TreeRec, a news recommendation framework based on user profile tree (UP-Tree), which is a novel framework combining content-based and collaborative filtering techniques. First, by exploitinga novel topic model namely UI-LDA, we obtain the representationvectorsfor news content in topic spaceas the fundamentalbridgeto associate user interests with news topics. Next, we design a decision tree with a dynamically changeable structure to construct a user interest profile from his feedback. Furthermore, we present a clustering based multidimensional similarity computation method to select the nearest neighbor of UP-Treeefficiently. We also provide a Map-Reduce framework based implementation that enables scaling our solution to real-world news recommendation problems.We conducted several experiments compared to the state-of-the-art approaches on real-world datasets and the experimental results demonstrate that our approach significantly improves the accuracy and effectiveness in news recommendation.
181Scopus© Citations 10 - PublicationConvolutional Matrix Factorization for Recommendation ExplanationIn this paper, we introduce a novel recommendation model, which harnesses a convolutional neural network to mine meaningful information from customer reviews, and integrates it with matrix factorization algorithm seamlessly. It is a valid method to improve the transparency of CF algorithms.
513Scopus© Citations 3 - PublicationMining Experiential Product Cases(2013-07-11)
; ; ; Case-based reasoning (CBR) attempts to reuse past experiences to solve new problems. CBR ideas are commonplace in recommendation systems, which rely on the similarity between product queries and a case base of product cases. But, the relationship between CBR and many of these recommenders can be tenuous: the idea that product cases made up of static meta-data type features are experiential is a stretch; unless one views the type of case descriptions used by collaborative filtering (user ratings across products) as experiential. Here we explore and evaluate how to automatically generate product cases from user-generated reviews to produce cases that are based on genuine user experiences for use in a case-based product recommendation system.200 - PublicationContext-Aware Co-attention Neural Network for Service RecommendationsContext-aware recommender systems are able to produce more accurate recommendations by harnessing contextual information, such as consuming time and location. Further, user reviews as an important information resource, providing valuable information about users' preferences, items' aspects, and implicit contextual features, could be used to enhance the embeddings of users, items, and contexts. However, few works attempt to incorporate these two types of information, i.e., contexts and reviews, into their models. Recent state-of-the-art context-aware methods only characterize relations between two types of entities among users, items and contexts, which may be insufficient, as the final prediction is closely related to all the three types of entities. In this paper, we propose a novel model, named Context-aware Co-Attention Neural Network (CCANN), to dynamically infer relations between contexts and users/items, and subsequently to model the degree of matching between users' contextual preferences and items' context-aware aspects via co-attention mechanism. To better leverage the information from reviews, we propose an embedding method, named Entity2Vec, to jointly learn embeddings of different entities (users, items and contexts) with words in a textual review. Experimental results, on three datasets composed of millions of review records crawled from TripAdvisor, demonstrate that our CCANN significantly outperforms state-of-the-art recommendation methods, and Entity2Vec can further boost the model's performance.
221Scopus© Citations 7 - PublicationThe Reviewer's Assistant: Recommending Topics to Writers by Association Rule Mining and Case-base Reasoning(2012-09-17)
; ; ; Today, online reviews for products and services have become an important class of user-generated content and they play a valuable role for countless online businesses by helping to convert casual browsers into informed and satisfied buyers. As users gravitate towards sites that offer insightful and objective reviews, the ability to source helpful reviews from a community of users is increasingly important. In this extended abstract we describe the Reviewer’s Assistant, a case-based reasoning inspired recommender system designed to help people to write more helpful reviews on sites such as Amazon and TripAdvisor. In particular, we describe two approaches to helping users during the review writing process and evaluate each as part of a blind live-user study. Our results point to high levels of user satisfaction and improved review quality compared to a control-set of Amazon reviews.199