A5087351600- Recommender Systems and Techniques
- Topic Modeling
- FinTech, Crowdfunding, Digital Finance
- Advanced Text Analysis Techniques
- Expert finding and Q&A systems
- Artificial Intelligence in Law
- Microfinance and Financial Inclusion
- Multimodal Machine Learning Applications
- Spam and Phishing Detection
- Law, Economics, and Judicial Systems
- Educational Technology and Assessment
- Software Engineering Research
- Information Retrieval and Search Behavior
- Teaching and Learning Programming
- Sentiment Analysis and Opinion Mining
- Domain Adaptation and Few-Shot Learning
- Software Testing and Debugging Techniques
- Text and Document Classification Technologies
- Legal Education and Practice Innovations
- Advanced Graph Neural Networks
- Online Learning and Analytics
- Natural Language Processing Techniques
- Advanced Bandit Algorithms Research
- Intelligent Tutoring Systems and Adaptive Learning
- Stochastic Gradient Optimization Techniques
University of Science and Technology of China
2019-2025
Huawei Technologies (China)
2022-2023
Recent years, Chinese text classification has attracted more and research attention. However, most existing techniques which specifically aim at English materials may lose effectiveness on this task due to the huge difference between English. Actually, as a special kind of hieroglyphics, characters radicals are semantically useful but still unexplored in classification. To that end, paper, we first analyze motives using multiple granularity features represent by inspecting characteristics...
Legal Judgment Prediction is a fundamental task in legal intelligence of the civil law system, which aims to automatically predict judgment results multiple subtasks, such as charge, article, and term penalty prediction. Existing studies mainly focus on impact entire fact description all subtasks. They ignore practical judicial scenario, where judges adopt circumstances crime (i.e., various parts fact) decide results. To this end, paper, we propose circumstance-aware prediction framework...
Online donation-based crowdfunding has brought new life to charity by soliciting small monetary contributions from crowd donors help others in trouble or with dreams. However, a crucial issue for platforms as well traditional charities is the problem of high donor attrition, i.e., many donate only once very few times within rather short lifecycle and then leave. Thus, it an urgent task analyze factors further predict behaviors. Especially, we focus on two types behavioral events, e.g.,...
In recommender systems, the cold-start problem is a critical issue. To alleviate this problem, an emerging direction adopts meta-learning frameworks and achieves success. Most existing works aim to learn globally shared prior knowledge across all users so that it can be quickly adapted new user with sparse interactions. However, may inadequate discern users’ complicated behaviors causes poor generalization. Therefore, we argue should locally by similar preferences who recognized social...
Massive legal documents have promoted the application of intelligence. Among them, Legal Judgment Prediction (LJP) has emerged as a critical task, garnering significant attention. LJP aims to predict judgment results for multiple subtasks, including charges, law articles, and terms penalty. Existing studies primarily focus on utilizing entire factual description produce results, overlooking practical judicial scenario where judges consider various crime circumstances decide verdicts...
Crowdfunding is an emerging mechanism for entrepreneurs or individuals to solicit funding from the public their creative ideas. However, in these platforms, quite a large proportion of campaigns (projects) fail raise enough money backers’ supports by declared expiration date. Actually, it very urgent predict exact success time campaigns. But this problem has not been well explored due series domain and technical challenges. In paper, we notice implicit factor distribution backing behaviors...
News recommender systems have become an effective manner to help users make decisions by suggesting the potential news that may click and read, which has shown proliferation nowadays. Many representative algorithms made great efforts discover users’ preferences from histories for triggering recommendations. However, there exist some limitations due following two main issues. First, they mainly rely on sufficient user data, cannot well capture temporal interests with very limited records....
Computerized Adaptive Testing (CAT) is a promising testing mode in personalized online education (e.g., GRE), which aims at measuring student's proficiency accurately and reducing test length. The "adaptive" reflected its selection algorithm that can retrieve best-suited questions for student based on his/her estimated each step. Although there are many sophisticated algorithms improving CAT's effectiveness, they restricted perturbed by the accuracy of current estimate, thus lacking...
Automatically generating controllable and diverse mathematical word problems (MWPs) which conform to equations topics is a crucial task in information retrieval natural language generation. Recent deep learning models mainly focus on improving the problem readability but overlook logic coherence, tends generate unsolvable problems. In this paper, we draw inspiration from human problem-designing process propose Mathematical structure Planning Knowledge enhanced Generation model (MaPKG),...
In the area of community question answering (CQA), answer selection and ranking are two tasks which applied to help users quickly access valuable answers. Existing solutions mainly exploit syntactic or semantic correlation between a its related answers (Q&A), where multi-facet domain effects in CQA still underexplored. this paper, we propose unified model, Enhanced Attentive Recurrent Neural Network (EARNN), for both by taking full advantages Q&A semantics (i.e., topic timeliness)....