Is this you? Claim this profile to correct it, add a bio and choose the work people see first.
Claim this profileWorks370 from public data
- Explainable AI: A Brief Survey on History, Research Areas, Approaches and Challenges693
The history of Explainable AI is introduced, starting from expert systems and traditional machine learning approaches to the latest progress in the context of modern deep learning, and the major research areas and the state-of-art approaches in recent years are described.
- Style Transfer in Text: Exploration and Evaluation550
This work proposes two novel evaluation metrics that measure two aspects of style transfer: transfer strength and content preservation, and shows that the proposed content preservation metric is highly correlate to human judgments.
- Multi-grained Attention Network for Aspect-Level Sentiment Classification520
Experimental results show that the multi-grained attention network consistently outperforms the state-of-the-art methods on all three datasets, and the effectiveness of aspect alignment loss indicates the aspect-level interactions can bring extra useful information and further improve the performance.
- Plan-and-Write: Towards Better Automatic Storytelling515
Experiments show that with explicit storyline planning, the generated stories are more diverse, coherent, and on topic than those generated without creating a full plan, according to both automatic and human evaluations.
- Semantic Relation Classification via Convolutional Neural Networks with Simple Negative Sampling304
This paper proposes to learn more robust relation representations from shortest dependency paths through a convolution neural network, takes the relation directionality into account and proposes a straightforward negative sampling strategy to improve the assignment of subjects and objects.
- Question Answering on Freebase via Relation Extraction and Textual Evidence303
This work first presents a neural network based relation extractor to retrieve the candidate answers from Freebase, and then infer over Wikipedia to validate these answers, a substantial improvement over the state-of-the-art.
- gStore302
This paper proposes a graph-based approach to store and query RDF data, which can answer exact SPARQL queries and queries with wildcards in a uniform manner, and proposes an effective maintenance algorithm to handle online updates over RDF repositories.
- Natural language question answering over RDF274
A semantic query graph is proposed to model the query intention in the natural language question in a structural way, based on which, RDF Q/A is reduced to subgraph matching problem and resolves the ambiguity of natural language questions at the time when matches of query are found.
- 271
- RUBER: An Unsupervised Method for Automatic Evaluation of Open-Domain Dialog Systems228
RUBER is proposed, a Referenced metric and Unreferenced metrics Blended Evaluation Routine, which evaluates a reply by taking into consideration both a groundtruth reply and a query (previous user-issued utterance).
- 208
- Knowledge-Grounded Dialogue Generation with Pre-trained Language Models206
Empirical results indicate that the proposed response generation defined by a pre-trained language model with a knowledge selection module and an unsupervised approach to jointly optimizing knowledge selection and response generation with unlabeled dialogues can significantly outperform state-of-the-art methods in both automatic evaluation and human judgment.
- Jointly Learning Entity and Relation Representations for Entity Alignment190
This paper presents a novel joint learning framework for entity alignment that is a Graph Convolutional Network (GCN) based framework for learning both entity and relation representations and incorporates the relation approximation into entities to iteratively learn better representations for both.
- Get The Point of My Utterance! Learning Towards Effective Responses with Multi-Head Attention Mechanism165
A novel Multi-Head Attention Mechanism (MHAM) for generative dialog systems, which aims at capturing multiple semantic aspects from the user utterance is proposed, and a regularizer is formulated to force different attention heads to concentrate on certain aspects.
- OVERCOMING CATASTROPHIC FORGETTING FOR CONTINUAL LEARNING VIA MODEL ADAPTATION155
This paper proposes a very different approach, called Parameter Generation and Model Adaptation (PGMA), to dealing with the problem of catastrophic forgetting in standard neural network architectures.
- Neighborhood Matching Network for Entity Alignment151
Neighborhood Matching Network (NMN), a novel entity alignment framework for tackling the structural heterogeneity challenge, can well estimate the neighborhood similarity in more tough cases and significantly outperforms 12 previous state-of-the-art methods.
- Multi-Representation Fusion Network for Multi-Turn Response Selection in Retrieval-Based Chatbots142
This work proposes a multi-representation fusion network where the representations can be fused into matching at an early stage, at an intermediate stage, or at the last stage, and demonstrates the effect of each representation to matching, which sheds light on how to select them in practical systems.
- How to Make Context More Useful? An Empirical Study on Context-Aware Neural Conversational Models133
This paper conducts an empirical study to compare various models and investigate the effect of context information in dialog systems, and proposes a variant that explicitly weights context vectors by context-query relevance, outperforming the other baselines.
- One Time of Interaction May Not Be Enough: Go Deep with an Interaction-over-Interaction Network for Response Selection in Dialogues129
Evaluation results on three benchmark data sets indicate that IoI can significantly outperform state-of-the-art methods in terms of various matching metrics and unveil how the depth of interaction affects the performance of IoI.
- Processing SPARQL queries over distributed RDF graphs124
This work proposes techniques for processing SPARQL queries over a large RDF graph in a distributed environment using a “partial evaluation and assembly” framework that is superior to the state-of-the-art methods in both the system’s performance and scalability.
- A Convolution BiLSTM Neural Network Model for Chinese Event Extraction122
This paper first comes up with the language specific issue in Chinese event extraction, and then proposes a convolution bidirectional L STM neural network that combines LSTM and CNN to capture both sentence-level and lexical information without any hand-craft features.
- An Ensemble of Retrieval-Based and Generation-Based Human-Computer Conversation Systems112
Experimental results show that a novel ensemble of retrieval-based and generation-based conversation system outperforms each single module by a large margin.
- Semantic SPARQL similarity search over RDF knowledge graphs108
This paper proposes an effective framework to access the RDF repository even if users have no full knowledge of the underlying schema, and is the first to propose a novel similarity measure, semantic graph edit distance, to improve the efficiency performance.
- VideoHallucer: Evaluating Intrinsic and Extrinsic Hallucinations in Large Video-Language Models99
This work introduces VideoHallucer, the first comprehensive benchmark for hallucination detection in large video-language models (LVLMs), which categorizes hallucinations into two main types: intrinsic and extrinsic, offering further subcategories for detailed analysis.
- Answering Natural Language Questions via Phrasal Semantic Parsing99
An efficient pipeline framework to model a user’s query intention as a phrase level dependency DAG which is then instantiated regarding a specific KB to construct the final structured query is proposed.
- VMSMO: Learning to Generate Multimodal Summary for Video-based News Articles96
A Dual-Interaction-based Multimodal Summarizer (DIMS), consisting of a dual interaction module and multimodal generator that handles the semantic relationship between news text and video from a high level.
- Learning with Noise: Enhance Distantly Supervised Relation Extraction with Dynamic Transition Matrix94
It is shown that the dynamic transition matrix can effectively characterize the noise in the training data built by distant supervision and can be effectively trained using a novel curriculum learning based method without any direct supervision about the noise.
- Towards Implicit Content-Introducing for Generative Short-Text Conversation Systems92
This paper proposes an implicit content-introducing method which incorporates additional information into the Seq2Seq model in a flexible way and fuse the general decoding and the auxiliary cue word information through the proposed hierarchical gated fusion unit.
- Product-Aware Answer Generation in E-Commerce Question-Answering90
The task of product-aware answer generation, which tends to generate an accurate and complete answer from large-scale unlabeled e-commerce reviews and product attributes is proposed, which is composed of three components: a question-aware review representation module, a key-value memory network encoding attributes, and a recurrent neural network as a sequence generator.
- 90
- An overview of kernel alignment and its applications89
The basic idea of kernel alignment and its theoretical properties, as well as the extensions and improvements for specific learning problems, are introduced and the typical applications, including kernel parameter tuning, multiple kernel learning, spectral kernel learning and feature selection and extraction are reviewed.
- GSN: A Graph-Structured Network for Multi-Party Dialogues86
The core of GSN is a graph-based encoder that can model the information flow along the graph-structured dialogues (two-party sequential dialogues are a special case) and Experimental results show that GSN significantly outperforms existing sequence-based models.
- CharacterChat: Learning towards Conversational AI with Personalized Social Support83
CharacterChat is presented, a comprehensive S2Conv system, which includes a conversational model driven by personas and memories, along with an interpersonal matching plugin model that dispatches the optimal supporters from the MBTI-1024 Bank for individuals with specific personas.
- Abstractive Text Summarization by Incorporating Reader Comments80
The task of reader-aware abstractive summary generation is proposed, which utilizes the reader comments to help the model produce better summary about the main aspect.
- 78
- 78
- Learning an Effective Context-Response Matching Model with Self-Supervised Tasks for Retrieval-based Dialogues76
This paper proposes learning a context-response matching model with auxiliary self-supervised tasks designed for the dialogue data based on pre-trained language models and jointly train the PLM-based response selection model with these auxiliary tasks in a multi-task manner.
- Interview Choice Reveals Your Preference on the Market72
A novel matching network with preference modeled is proposed to explore the latent preference given the history of all interviewed candidates for a job posting and thehistory of all job applications for a particular talent to improve job-resume matching on the job market.
- Multi-Granularity Structural Knowledge Distillation for Language Model Compression70
A novel knowledge distillation framework that gathers intermediate representations from multiple semantic granularities and forms the knowledge as more sophisticated structural relations specified as the pair-wise interactions and the triplet-wise geometric angles based on multi-granularity representations is presented.
- Capturing Relations between Scientific Papers: An Abstractive Model for Related Work Section Generation70
A relation-aware multi-document encoder that relates one document to another according to their content dependency in a relation graph that generates an abstractive related work from the given multiple scientific papers in the same research area is proposed.
- Generating Classical Chinese Poems via Conditional Variational Autoencoder and Adversarial Training67
Experimental results on a large poetry corpus confirm the validity and effectiveness of the proposed conditional variational autoencoder with adversarial training for classical Chinese poem generation, where its automatic and human evaluation scores outperform existing models.
- Efficient simrank-based similarity join over large graphs65
This paper adopts "SimRank" to evaluate the similarity of two vertices in a large graph because of its generality, and extends the technique to the partition-based framework.
- Adaptive Orthogonal Projection for Batch and Online Continual Learning62
A novel method is proposed, called AOP (Adaptive Orthogonal Projection), to resolve an issue that was discovered in the mathematical derivation of this approach and results in significant accuracy gains in empirical evaluations in both the batch and online continual learning settings without saving any previous training data as in replay-based methods.
- Towards Effective and Interpretable Person-Job Fitting62
The proposed IPJF model leverages deep interactive representation learning to automatically learn the interdependence between a resume and job requirements without relying on a clear list of job seeker's abilities, and deploys the optimizing problem as a learning to rank problem.
- Graph similarity search with edit distance constraint in large graph databases62
This paper derives a lower bound, branch-based bound, which can greatly reduce the search space of the graph similarity search, and proposes a tree index structure, namely b-tree, to facilitate effective pruning and efficient query processing.
- How to Build Templates for RDF Question/Answering61
This work is the first to propose a join approach for template generation in RDF Q/A, and proposes several structural and probability pruning techniques to speed up joining.
- Three Sentences Are All You Need: Local Path Enhanced Document Relation Extraction60
This paper presents an embarrassingly simple but effective method to heuristically select evidence sentences for document-level RE, which can be easily combined with BiLSTM to achieve good performance on benchmark datasets, even better than fancy graph neural network based methods.
- Natural Language Understanding and Intelligent Applications60
A novel word sense disambiguation (WSD) discriminative model is proposed which integrates structural context with the local context to handle long distance sense dependency and multireference lexicon dependency within the sentence.
- Why Machine Reading Comprehension Models Learn Shortcuts?59
It is argued that larger proportion of shortcut questions in training data make models rely on shortcut tricks excessively, and two new methods are proposed to quantitatively analyze the learning difficulty regarding shortcut and challenging questions, and revealing the inherent learning mechanism behind the different performance between the two kinds of questions.
- Scale Up Event Extraction Learning via Automatic Training Data Generation57
This work develops an automatic approach for generating training data for event extraction that allows to scale up event extraction training instances from thousands to hundreds of thousands, and it does this at a much lower cost than a manual approach.
- Recommending academic papers via users' reading purposes57
This paper proposes to satisfy user-specific reading purposes by recommending the most problem- related papers or solution-related papers to users separately, and demonstrates that this approach outperforms a traditional similarity based model and can provide highly relevant paper recommendations according to different reading purposes for researchers.
- Predictive Adversarial Learning from Positive and Unlabeled Data56
Evaluation of the proposed novel Predictive Adversarial Networks (PAN) based on GAN shows that PAN outperforms state-of-the-art PU learning methods and also a direct adaptation of GAN for PU learning.
- Hybrid Question Answering over Knowledge Base and Free Text56
This paper presents a hybrid question answering (hybrid-QA) system which exploits both structured knowledge base and free text to answer a question, and develops an integer linear program (ILP) model to infer on these candidates and provide a globally optimal solution.
- Keyword Search on RDF Graphs - A Query Graph Assembly Approach55
This paper formally defines the query graph assembly (QGA) problem, design some heuristic lower bounds and propose a bipartite graph matching-based best-first search algorithm, which has good scalability in large RDF graphs.
- Semi-supervised Text Style Transfer: Cross Projection in Latent Space53
A semi-supervised text style transfer model that combines the small-scale parallel data with the large-scale nonparallel data is proposed and a projection function between the latent space of different styles is introduced.
- Meaningful Answer Generation of E-Commerce Question-Answering50
A novel generative neural model, called the Meaningful Product Answer Generator (MPAG), which alleviates the safe answer problem by taking product reviews, product attributes, and a prototype answer into consideration and achieves state-of-the-art performance in terms of both automatic metrics and human evaluations.
- Learning towards Abstractive Timeline Summarization49
This paper proposes a memory-based timeline summarization model (MTS), which tends to concisely paraphrase the information in the time-stamped events and achieves the state-of-the-art performance in terms of both automatic and human evaluations.
- From Standard Summarization to New Tasks and Beyond: Summarization with Manifold Information48
This paper focuses on the survey of these new summarization tasks and approaches in the real-world application of text summarization algorithms.
- A Character-Centric Neural Model for Automated Story Generation48
This work proposes a character-centric neural storytelling model, where each part of a story is conditioned on a given character and corresponded context environment, and explicitly capture the character information and the relations between plots and characters to improve explainability and consistency.
- Lattice CNNs for Matching Based Chinese Question Answering48
Experimental results show that the LCNs models can significantly outperform the state-of-the-art matching models and strong baselines by taking advantages of better ability to distill rich but discriminative information from the word lattice input.
- Joint Learning of Response Ranking and Next Utterance Suggestion in Human-Computer Conversation System48
A novel Dual-LSTM Chain Model is proposed to couple response ranking and next utterance suggestion simultaneously simultaneously and it is assumed that the new conversation mode is more proactive and keeps user engaging.
- Are Training Samples Correlated? Learning to Generate Dialogue Responses with Multiple References47
Experimental results show that the proposed model can effectively improve the quality of response and outperform existing neural dialogue models on both automatic and human evaluations.
- Time Constrained Continuous Subgraph Search Over Streaming Graphs47
This paper proposes a solution to efficiently answer subgraph search, introduces optimizations to greatly reduce the space cost, and design concurrency management to improve system throughput.
- Modeling Personalization in Continuous Space for Response Generation via Augmented Wasserstein Autoencoders46
Experimental results on a large-scale real-world dataset confirm the superiority of the proposed model for generating informative and personalized responses, where both automatic and human evaluations outperform state-of-the-art models.
- Lattice-BERT: Leveraging Multi-Granularity Representations in Chinese Pre-trained Language Models45
This work proposes a novel pre-training paradigm for Chinese — Lattice-BERT, which explicitly incorporates word representations along with characters, thus can model a sentence in a multi-granularity manner and achieves new state-of-the-art among base-size models on the CLUE benchmarks.
- Dynamic Skyline Queries in Large Graphs45
This paper proposes a novel pruning rule based on graph properties to derive the candidates for DSG-query, that are guaranteed not to introduce false negatives, and employs a filter-and-refine framework to speed up the query processing.
- A Document-grounded Matching Network for Response Selection in Retrieval-based Chatbots44
A document-grounded matching network for response selection that can power a knowledge-aware retrieval-based chatbot system that can significantly improve upon state-of-the-art methods and at the same time enjoys good interpretability.
- Marrying Up Regular Expressions with Neural Networks: A Case Study for Spoken Language Understanding44
Novel methods to exploit the rich expressiveness of REs at different levels within a NN are developed, showing that the combination significantly enhances the learning effectiveness when a small number of training examples are available.
- 44
- Efficient processing of label-constraint reachability queries in large graphs44
This paper proposes the path-label transitive closure method to answer LCR queries in large graphs, and proves that finding the optimal graph partition to minimize query processing cost is a NP-hard problem.
- 43
- Everything Has a Cause: Leveraging Causal Inference in Legal Text Analysis42
Experimental results show that GCI can capture the nuance from fact descriptions among multiple confusing charges and provide explainable discrimination, especially in few-shot settings, and that the causal knowledge contained inGCI can be effectively injected into powerful neural networks for better performance and interpretability.
- Learning to Respond with Stickers: A Framework of Unifying Multi-Modality in Multi-Turn Dialog42
A sticker response selector (SRS) model that employs a convolutional based sticker image encoder and a self-attention based multi-turn dialog encoder to obtain the representation of stickers and utterances and achieves the state-of-the-art performance for all commonly-used metrics.
- Iterative Document Representation Learning Towards Summarization with Polishing42
This paper introduces Iterative Text Summarization (ITS), an iteration-based model for supervised extractive text summarization inspired by the observation that it is often necessary for a human to read an article multiple times in order to fully understand and summarize its contents.
- Insufficient Data Can Also Rock! Learning to Converse Using Smaller Data with Augmentation41
A novel generative model is proposed to augment existing data, where the conditional variational autoencoder is employed as the generator to output more training data with diversified expressions to improve the performance of neural dialogue models on the condition of insufficient data.
- Dealing with Cross-Task Class Discrimination in Online Continual Learning36
A novel optimization objective with a gradient-based adaptive method is proposed to dynamically deal with the problem of cross-task class discrimination in the online CL process and experimental results show that the new method achieves much better results in online CL.
- Coupled Context Modeling for Deep Chit-Chat36
A novel context modeling framework with end-to-end neural networks for human-computer conversational systems that will search for appropriate responses, conduct relevance ranking using contexts information, and then output the highly relevant result.
- Synergistic Interplay between Search and Large Language Models for Information Retrieval35
InteR, a novel framework that facilitates information refinement through synergy between RMs and LLMs, allows RMs to expand knowledge in queries using LLM-generated knowledge collections and enables LLMs to enhance prompt formulation using retrieved documents, leading to more accurate retrieval.
- Dialogue History Matters! Personalized Response Selection in Multi-Turn Retrieval-Based Chatbots35
This work incorporates user-specific dialogue history into the response selection and proposes a personalized hybrid matching network (PHMN), which significantly outperforms several strong models by combining personalized attention, wording behaviors, and hybrid representation learning.
- Learning to Update Knowledge Graphs by Reading News35
GUpdater is a novel neural network method built upon graph neural networks (GNNs) with a text-based attention mechanism to guide the updating message passing through the KG structures, and can effectively broadcast the news information to theKG structures and perform necessary link-adding or link-deleting operations to ensure the KGs up-to-date according to news snippets.
- Sampling Matters! An Empirical Study of Negative Sampling Strategies for Learning of Matching Models in Retrieval-based Dialogue Systems35
Empirical studies indicate that compared with the widely used random sampling strategy, although the first two strategies lead to performance drop, the latter two ones can bring consistent improvement to the performance of all the models on both benchmarks.
- Two-stage multiple kernel learning with multiclass kernel polarization35
This paper proposes a simple but effective multiclass MKL method by a two-stage strategy, in which the first stage finds the kernel weights to combine the kernels, and the second stage trains a standard multiclass support vector machine (SVM).
- Combining Curriculum Learning and Knowledge Distillation for Dialogue Generation33
A hierarchical curriculum is arranged for the student model under the guidance from the teacher model based on curriculum learning and knowledge distillation for dialogue generation models, where curriculum learning can help know distillation from data and model aspects.
- Who Is Speaking to Whom? Learning to Identify Utterance Addressee in Multi-Party Conversations33
This paper introduces a novel who-to-whom (W2W) model which models users and utterances in the session jointly in an interactive way and demonstrates that the model outperforms baselines with consistent improvements.
- 33
- Answering pattern match queries in large graph databases via graph embedding33
This paper addresses distance-based pattern match queries over a large data graph G by adopting a filter-and-refine framework and finding a set of candidate matches by a graph embedding technique and then evaluating these to find the exact matches.
- Modeling discourse cohesion for discourse parsing via memory network32
This paper proposes a new transition-based discourse parser that makes use of memory networks to take discourse cohesion into account, and the automatically captured discourse cohesion benefits discourse parsing, especially for long span scenarios.
- Selection and Generation: Learning towards Multi-Product Advertisement Post Generation31
A novel end-to-end model named S-MG Net is proposed to generate the advertisement (AD) post and achieves impressive performance in terms of both automatic metrics as well as human evaluations.
- How to Write Summaries with Patterns? Learning towards Abstractive Summarization through Prototype Editing31
Extensive experiments conducted on a large-scale real-world text summarization dataset show that PESG achieves the state-of-the-art performance in terms of both automatic metrics and human evaluations.
- Language Models can be Logical Solvers30
LoGiPT is a novel language model that directly emulates the reasoning processes of logical solvers and bypasses the parsing errors by learning to strict adherence to solver syntax and grammar and outperforms state-of-the-art solver-augmented LMs and few-shot prompting methods on competitive LLMs.
- Towards a Neural Conversation Model With Diversity Net Using Determinantal Point Processes30
A novel framework which organically combines seq2seq model with Determinantal Point Processes is proposed, which achieves high quality in generated reply and significantly improves the diversity among them.
- 29
- A Novel Joint Framework for Multiple Chinese Events Extraction29
A novel joint framework for multiple Chinese event extraction is proposed which jointly performs predictions for event triggers and arguments based on shared feature representations from pre-trained language model.
- Learning a Matching Model with Co-teaching for Multi-turn Response Selection in Retrieval-based Dialogue Systems28
To learn a robust matching model from noisy training data, a general co-teaching framework with three specific teaching strategies that cover both teaching with loss functions and teaching with data curriculum is proposed.
- Learning to Converse with Noisy Data: Generation with Calibration28
This paper proposes a generation with calibration framework, that allows high- quality data to have more influences on the generation model and reduces the effect of noisy data.
- Draft and Edit: Automatic Storytelling Through Multi-Pass Hierarchical Conditional Variational Autoencoder27
A multi-pass hierarchical conditional variational autoencoder model is proposed to overcome the challenges and limitations in existing automatic storytelling models and yields substantial improvement over the existing state-of-the-art approaches.
- 26
- Cross-Lingual Low-Resource Set-to-Description Retrieval for Global E-Commerce26
A new task of cross-lingual set-to-description retrieval in cross-border e-commerce, which involves matching product attribute sets in the source language with persuasive product descriptions in the target language is explored, and a novel cross- Lingual matching network (CLMN) is proposed with the enhancement of context-dependent cross-lingsual mapping upon the pre-trained monolingual BERT representations.
- Extracting 5W1H Event Semantic Elements from Chinese Online News26
A verb-driven approach to extract 5W1H (Who, What, Whom, When, Where and How) event semantic information from Chinese online news by considering valency of a Chinese verb and building a prototype system named Chinese News Fact Extractor (CNFE).
- Language Models can be Deductive Solvers25
This paper introduces L O G I PT, a novel language model that directly internal-izes and emulates the reasoning processes of logical solvers and avoids parsing errors by learning strict adherence to solver syntax and grammar.
- 25
- Stick to the Facts: Learning towards a Fidelity-oriented E-Commerce Product Description Generation25
This work proposes a model named Fidelity-oriented Product Description Generator (FPDG), which takes the entity label of each word into account, since the product attribute information is always conveyed by entity words.
- 24
- Follow the Timeline! Generating an Abstractive and Extractive Timeline Summary in Chronological Order24
The Unified Timeline Summarizer is proposed, which can generate abstractive and extractive timeline summaries in time order and achieves state-of-the-art performance in terms of both automatic and human evaluations.
- Using Conditional Functional Dependency to Discover Abnormal Data in RDF Graphs24
This paper innovatively defines Graph-based Conditional Functional Dependency (GCFD) that can represent the attribute value and semantic structure dependencies of RDF data in a uniform manner and proposes an efficient framework and some novel pruning rules to discover GCFD in large RDF graphs.
- Smarter Response with Proactive Suggestion: A New Generative Neural Conversation Paradigm23
This paper proposes a new paradigm for neural generative conversations: smarter response with a suggestion is provided given the query and proposes a novel integrated model to handle both the response generation and the suggestion generation.
- Open Domain Question Answering System Based on Knowledge Base23
A SPE (subject predicate extraction) algorithm which can automatically extract a subject-predicate pair from a simple question and translate it to a KB query is proposed.
- Towards Context-Aware Code Comment Generation22
A novel approach to automatically generate code comments at a function level by targeting object-oriented programming languages and designs a novel learning framework based on the bidirectional gated recurrent unit and a graph attention network with a pointer mechanism.
- Query Workload-based RDF Graph Fragmentation and Allocation22
This paper proposes a method to explore the intrinsic similarities among the structures of queries in a workload for fragmentation and allocation, which aims to reduce the number of crossing matches and the communication cost during SPARQL query processing.
- Answering label-constraint reachability in large graphs22
This paper studies a variant of reachability queries, called label-constraint reachability (LCR) queries, and proves the superiority of this method by extensive experiments.
- 22
- 21
- The Style-Content Duality of Attractiveness: Learning to Write Eye-Catching Headlines via Disentanglement21
A Disentanglement-based Attractive Headline Generator (DAHG) that generates headline which captures the attractive content following the attractive style and takes the polished document as input to generate headline under the guidance of the attractive Style.
- Question Answering via Phrasal Semantic Parsing21
An efficient pipeline framework to model a user's query intention as a phrase level dependency DAG which is then instantiated regarding a specific KB to construct the final structured query is proposed.
- Top-k queries on RDF graphs21
An effective and efficient processing algorithm for top-k queries that consists of a novel tree-style index MS-tree and MS- tree-based filtering and pattern-matching functions and a dynamic variable selecting optimization is employed to accelerate the threshold decrease.
- S-store: An Engine for Large RDF Graph Integrating Spatial Information21
SPARQL, a variant of SPARQL language, is introduced for querying spatial information integrated RDF data and a novel index SS-tree is designed for evaluating the spatial queries and a search algorithm is proposed.
- EZInterviewer: To Improve Job Interview Performance with Mock Interview Generator20
A novel application named EZInterviewer is proposed, which aims to learn from the online interview data and provides mock interview services to the job seekers and achieves promising results to generate mock interviews.
- Learning to Write Stories with Thematic Consistency and Wording Novelty20
Experimental results indicate that the proposed cache-augmented conditional variational autoencoder model can generate stories with consistency and wording novelty, and outperforms existing models under both automatic metrics and human evaluations.
- Personalized News Recommendation Using Ontologies Harvested from the Web20
This paper proposes an Ontology Based Similarity Model (OBSM) to calculate the news-user similarity through collaboratively built ontological structures and compares its approach with other ontology-based baselines on both English and Chinese data sets.
- 19
- Dialogue Summarization with Static-Dynamic Structure Fusion Graph18
A Static-Dynamic graph-based Dialogue Summarization model (SDDS), which fuses prior knowledge from human expertise and adaptively learns the graph structure in an end-to-end learning fashion is proposed and the results verify the superiority of SDDS.
- Learning to Respond with Your Favorite Stickers18
This article proposes a Preference Enhanced Sticker Response Selector (PESRS) model, which learns the short-term and long-term dependency between all interaction results by a fusion network and dynamically fuses the user preference representation into the final sticker selection prediction.
- What If Bots Feel Moods?18
An emotion-aware transition network is proposed to model the dynamic emotional flow and enhance context-response matching in retrieval-based dialogue systems with learnt intrinsic emotion features through a multi-task learning framework and is applied to a real IoT application.
- A Chinese Question Answering System for Single-Relation Factoid Questions18
A question answering system which can automatically find the promised entities and predicates for single-relation questions by using deep convolutional neural networks to rerank the entity-predicate pairs.
- g -store: Querying Large Spatiotemporal RDF Graphs18
This paper formally defines the spatiotemporal information-integrated RDF data, introduces a spatiotsemporal query language that extends the SPARQL language with spatiotmporal assertions to query spatiotechnological information-Integrated R DF data, and designs a novel index and the corresponding query algorithm.
- 17
- Efficient Subgraph Skyline Search Over Large Graphs17
This paper proposes a hybrid feature encoding incorporating both structural and numeric features and presents some optimizations based on partitioning strategy to improve the efficiency of subgraph skyline search.
- 16
- Learning Multi-turn Response Selection in Grounded Dialogues with Reinforced Knowledge and Context Distillation16
A new knowledge-grounded dialogue model based on PLMs is proposed, where a knowledge selector and a context selector are designed for filtering out irrelevant knowledge sentences and redundant dialogue history, respectively, and results indicate that the model can significantly outperform the state-of-the-art methods.
- 16
- 16
- Reasoning in Dialog: Improving Response Generation by Context Reading Comprehension15
This paper proposes a joint framework that unifies these two tasks, sharing the same encoder to extract the common and task-invariant features with different decoders to learn task-specific features, and augments the Transformer architecture with a memory updater, designed to selectively store and update the history dialog information so as to support downstream tasks.
- Enhancing the Open-Domain Dialogue Evaluation in Latent Space15
This paper proposes a self-supervised setting to obtain a smooth latent space that can both capture discourse-level context information and implicitly model more references in latent space and demonstrates the superiority of this method for open-domain dialogue evaluation.
- How does Truth Evolve into Fake News? An Empirical Study of Fake News Evolution15
The Fake News Evolution (FNE) dataset is presented: a new dataset tracking the fake news evolution process, composed of 950 paired data that observe the features during the evolution and they are the disinformation techniques, text similarity, top 10 keywords, classification accuracy, parts of speech, and sentiment properties.
- Context-to-Session Matching: Utilizing Whole Session for Response Selection in Information-Seeking Dialogue Systems15
The response and its context as a whole session is considered and the task of matching the query's context with the sessions is explored and the proposed context-to-session method outperforms the strong baselines significantly.
- Efficient SimRank-Based Similarity Join15
This article adopts “SimRank” to evaluate the similarity between two vertices in a large graph because of its generality, and proposes an efficient method without building the vertex-pair graph to find the h-go cover + vertex pairs.
- 15
- SQBC: An efficient subgraph matching method over large and dense graphs15
A novel Subgraph Query technique Based on Clique feature, called SQBC, is presented, which integrates the carefully designed clique encoding with the existing vertex encoding as the basic index unit to reduce the search space.
- 15
- Learning to Plan with Natural Language14
The Learning to Program (LP) method is proposed to ask LLMs themselves to learn natural language programs from the training dataset of complex tasks and then use the learned program to guide inference.
- Content Learning with Structure-Aware Writing: A Graph-Infused Dual Conditional Variational Autoencoder for Automatic Storytelling14
This work proposes a graph-infused dual conditional variational autoencoder model to capture multi-level intra-story structures by continuous variational latent variables and generate consistent stories through dual-infusion of story structure planning and content learning.
- 14
- Taxonomy Based Personalized News Recommendation: Novelty and Diversity14
The experimental results show that the coarse grained knowledge resources can help a content-based news recommender system provides accurate as well as user-oriented recommendations.
- Decouple knowledge from parameters for plug-and-play language modeling13
PlugLM is introduced, a pre-training model with differentiable plug-in memory (DPM) that can de-couple the knowledge storage from model parameters with an editable and scalable key-value memory and leverage knowledge in an explainable manner by knowledge retrieval in the DPM.
- Towards Personalized Review Summarization by Modeling Historical Reviews from Customer and Product Separately13
The Heterogeneous Historical Review aware Review Summarization Model (HHRRS) is proposed which separately models the two types of historical reviews with the rating information by a graph reasoning module with a contrastive loss.
- ProphetChat: Enhancing Dialogue Generation with Simulation of Future Conversation13
A novel dialogue generation framework named ProphetChat is proposed that utilizes the simulated dialogue futures in the inference phase to enhance response generation and demonstrates that ProphetChat can generate better responses over strong baselines, which validates the advantages of incorporating the simulatedDialogue futures.
- Multilingual COVID-QA: Learning towards Global Information Sharing via Web Question Answering in Multiple Languages13
A novel framework which incorporates (unsupervised) translation alignment to learn as pseudo-parallel data is proposed and the effectiveness of the proposed approach compared against a series of competitive baselines is demonstrated.
- 13
- Understanding Procedural Text using Interactive Entity Networks13
This paper proposes a novel Interactive Entity Network (IEN), which is a recurrent network with memory equipped cells for state tracking that outperforms state-of-the-art models by precisely capturing the interactions of multiple entities and explicitly leverage the relationship between entity interactions and subsequent state changes.
- Be Aware of the Hot Zone: A Warning System of Hazard Area Prediction to Intervene Novel Coronavirus COVID-19 Outbreak13
A system predicting hazard areas on the basis of confirmed infection cases with location information and achieving promising overall performance in terms of precision, recall, accuracy, F1 score, and AUC is built.
- One "Ruler" for All Languages: Multi-Lingual Dialogue Evaluation with Adversarial Multi-Task Learning13
Experiments show that the adversarial multi-task neural metric (ADVMT) achieves a high correlation with human annotation, which yields better performance than monolingual ones and various existing metrics.
- Diversifying Neural Conversation Model with Maximal Marginal Relevance.13
This paper proposes the MMR-BS decoding method, which incorporates MMR into the beam search (BS) process of seq2seq, and shows that the proposed model achieves the best performance among other comparison methods.
- Top-K Possible Shortest Path Query over a Large Uncertain Graph13
Lower and upper bounds for the probability that one path is the shortest path in an uncertain graph are developed and, based on which, a filter-and-refine framework is proposed to answer kSP queries.
- Learning Disentangled Representation via Domain Adaptation for Dialogue Summarization12
A disentangled representation method to reduce the deviation between data in different domains, where the input data is disentangling into domain-invariant and domain-specific representations, and two novel reconstruction strategies, namely backtracked and cross-track reconstructions, are proposed.
- 12
- 12
- 12
- Multi-query Optimization in Federated RDF Systems12
A heuristic query rewriting-based approach to share the common computation during evaluation of multiple queries while considering the cost of both query evaluation and data shipment and an efficient method to use the interconnection topology between RDF sources to filter out irrelevant sources and share thecommon computation of intermediate results joining.
- 12
- Learning to Organize a Bag of Words into Sentences with Neural Networks: An Empirical Study11
By running comprehensive comparisons, this paper quantitatively compare the ability of several representative neural models to organize sentences from a bag of words under three typical scenarios, and summarizes some empirical findings and challenges, which can shed light on future research on this line of work.
- 11
- QAEval: Mixture of Evaluators for Question-Answering Task Evaluation10
QAE-val, a hybrid framework combining rule-based reliability with LLM-based adaptability, enables accurate, cost-effective QA evaluation and outperforms models like GPT-4o and Claude-3, Experimental results show.
- HeteroQA10
This work proposes a question-aware heterogeneous graph transformer to incorporate the MIS in the user community to automatically generate the answer in Community Question Answering.
- 10
- Response Ranking with Multi-types of Deep Interactive Representations in Retrieval-based Dialogues10
The model can handle the relation of neighboring elements, phrasal pattern, and long-range dependencies during the representation and make a more accurate prediction through multiple layers of interactions between the context–response pair.
- Encoding implicit relation requirements for relation extraction: A joint inference approach10
Experimental results on both English and Chinese datasets show that the proposed framework can effectively utilize those two categories of global clues and resolve the disagreements among local predictions, thus improve various relation extractors when such clues are applicable to the datasets.
- Improved Discourse Parsing with Two-Step Neural Transition-Based Model10
A novel parsing model to analyze discourse in a two-step fashion with different feature representations to characterize intra sentence and inter sentence discourse structures, respectively is proposed.
- Knowledge Base Completion Using Matrix Factorization10
This paper treats KBs completion as a large matrix completion task and incorporates the similarity of relations between different KBs into matrix factorization model to take full advantage of the complementarity of various KBs, achieving significant accuracy improvements in binary relation prediction.
- Answering subgraph queries over massive disk resident graphs10
This paper focuses on subgraph query over a single large graph G, i.e., finding all embeddings of query Q in G and proposes a bitmap structure to index R2, which has the linear space complexity instead of exponential complexity in feature-based approaches.
- 9
- Your Career Path Matters in Person-Job Fit9
This paper proposes a new approach to person-job fit that characterizes the hidden preference derived from the job seeker's career path and proves that understanding the career path enables us to provide more appropriate career suggestions to job seekers.
- StyleChat: Learning Recitation-Augmented Memory in LLMs for Stylized Dialogue Generation9
Experimental results show that the proposed framework StyleChat outperforms all the baselines and helps to break the style boundary of LLMs, which has been carefully constructed with rigorous human-led quality control.
- Rethinking Task-Specific Knowledge Distillation: Contextualized Corpus as Better Textbook9
Surprisingly, contextualized corpus enables task-specific distillation from scratch without general distillation while maintaining comparable performance, making it more flexible to customize the student model with desired model size under various computation constraints.
- 9
- 9
- 9
- Follow the Timeline! Generating Abstractive and Extractive Timeline Summary in Chronological Order8
A Unified Timeline Summarizer (UTS) that can generate abstractive and extractive timeline summaries in time order is proposed and achieves state-of-the-art performance in terms of both automatic and human evaluations.
- 8
- 8
- Finding the Dominant Winning Ticket in Pre-Trained Language Models8
Strikingly, it is found that a dominant winning ticket that takes up 0.05% of the parameters can already achieve satisfactory performance, indicating that the PLM is significantly reducible during fine-tuning.
- Plan-CVAE: A Planning-Based Conditional Variational Autoencoder for Story Generation8
This paper proposes a planning-based conditional variational autoencoder, namely Plan-CVAE, which first plans a keyword sequence and then generates a story based on the keyword sequence, and outperforms state-of-the-art methods on both automatic metrics and human evaluations.
- Evaluating and Enhancing the Robustness of Retrieval-Based Dialogue Systems with Adversarial Examples8
With adversarial training using the generated adversarial examples, this paper significantly improves the robustness of retrieval-based dialogue systems and provides new insights to facilitate future work on building more robust dialogue systems.
- 7
- 7
- 7
- GNN-encoder: Learning a Dual-encoder Architecture via Graph Neural Networks for Dense Passage Retrieval7
A GNN-encoder model is proposed in which query (passage) information is fused into passage ( query) representations via graph neural networks that are constructed by queries and their top retrieved passages to achieve both efficiency and efficacy in passage retrieval.
- Learning to Express in Knowledge-Grounded Conversation7
This work proposes a segmentation-based generation model and optimize the model by a variational approach to discover the underlying pattern of knowledge expression in a response to generate responses in desired content style.
- 7
- 7
- 7
- 7
- 7
- 7
- 6
- AnaFig: A Human-Aligned Dataset for Scientific Figure Analysis6
An SFA dataset, AnaFig, is introduced, comprising 2,000 high-quality samples across 56 domains, facilitating the assessment of three critical capabilities of multimodal large language models (MLLMs): adherence to complex instructions, multimodal perception, and analytical summarization.
- RetriEVAL: Evaluating Text Generation with Contextualized Lexical Match6
The proposed RetriEVAL is an unsupervised automatic evaluation metric inspired by the concept of lexical match in information retrieval that leverages contextualized lexical matching to measure exact matches between identical tokens and dynamically matches different tokens based on their contextualized representations.
- 6
- 6
- 6
- 6
- 6
- 6
- 6
- When Less Is More: Using Less Context Information to Generate Better Utterances in Group Conversations6
It is observed that fewer contexts can lead to better performance by tackling the structure of group conversations and the model outperforms baselines.
- 6
- 6
- 6
- 6
- 6
- 6
- 5
- 5
- FAA: Fine-grained Attention Alignment for Cascade Document Ranking5
This work utilizes the attention activations over the passages from the ranker as fine-grained attention feedback to optimize the selector and fuse the relevance scores from the passage selector into theRanker to assist in calculating the cooperative matching representation.
- 5
- 5
- 5
- 5
- 5
- BioGen: Generating Biography Summary under Table Guidance on Wikipedia5
The TaGS (Ta ble-Guided Summarization) dataset is introduced, the first large-scale biography summarization dataset with tables, and several commonly used summarization methods are benchmarked on TaGS to inspire more exciting methods.
- Translation vs. Dialogue: A Comparative Analysis of Sequence-to-Sequence Modeling5
This paper studies the sequence-to-sequence (Seq2Seq) model in the contexts of two mainstream NLP tasks as they both use the seq2seq model to investigate how the two tasks are different and how their task difference results in major differences in the behaviors of the resulting translation and dialogue generation systems.
- 5
- 5
- 5
- 4
- 4
- 4
- 4
- 4
- 4
- 4
- Learning Sense Representation from Word Representation for Unsupervised Word Sense Disambiguation (Student Abstract)4
This work proposes SGM and MGM to learn sense representations for words in WordNet using the glosses and calculates the similarity between each sense of the ambiguous word and its context to select the sense with the highest similarity in the WSD task.
- Improving Matching Models with Hierarchical Contextualized Representations for Multi-turn Response Selection4
Experimental results indicate that the proposed hierarchical contextualized representations can bring significantly and consistently improvement to existing matching models for response selection.
- Improving Matching Models with Contextualized Word Representations for Multi-turn Response Selection in Retrieval-based Chatbots.4
Experimental results on two benchmark datasets indicate that the proposed contextualized word vectors can significantly and consistently improve the performance of existing matching models for response selection.
- 4
- 4
- 4
- PKUICST at TREC 2015 Microblog Track: Query-biased Adaptive Filtering in Real-time Microblog Stream.4
- 4
- 4
- 4
- 4
- 4
- 3
- Large Language Models Can Learn Representation in Natural Language3
This paper introduces the Learning to Represent with Natural Language method, which utilizes LLMs to optimize entity representations consisting of text patterns based on environmental feedback, and successfully learns human-understandable representations for classification tasks and API call tasks.
- 3
- 3
- 3
- 3
- 3
- 3
- 3
- Query-to-Session Matching: Do NOT Forget History and Future during Response Selection for Multi-Turn Dialogue Systems3
This paper is the first work to utilize both the response's history and future in the response selection task, and the experimental results show that the proposed model outperforms existing state-of-the-art methods by a large margin.
- 3
- 3
- 3
- 3
- 3
- 3
- 3
- 2
- 2
- 2
- 2
- 2
- 2
- 2
- 2
- 2
- 2
- 2
- 2
- Multi-Response Awareness for Retrieval-Based Conversations: Respond with Diversity via Dynamic Representation Learning2
A Respond-with-Diversity model augmented by the memory module interacting with both the query utterances and multiple candidate responses is proposed, and dynamic representations for the input queries conditioned on different response candidates are obtained.
- 2
- Query-bag Matching with Mutual Coverage for Information-seeking Conversations in E-commerce2
A query-bag matching model is proposed which mainly utilizes the mutual coverage between query and bag and measures the degree of the content in the query mentioned by the bag, and vice verse.
- 2
- 2
- 2
- 2
- 2
- 2
- 2
- 2
- 2
- 2
- Natural Language Processing and Chinese Computing: First CCF Conference, NLPCC 2012, Beijing, China, October 31-November 5, 2012. Proceedings2
Experimental results show that users’ historical behavior plays an important role in user preference modeling and the proposed method improves the final predication performance in the field of technical papers recommendation.
- 2
- SciChart: Visual Question Answering and Reasoning for Scientific Spectral Chart1
This work designs two tasks, basic question answering (BasicQA) and reasoning-based question answering (ReaQA), to evaluate the models' ability to directly extract information from charts, and understand the textual and visual information for reasoning.
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- Dimension-Prompts Boost Commonsense Consolidation–
A brand-new dimension-disentangled knowledge model (D2KM) learning paradigm with multiple sources is proposed that is, a generative language model with dimension-specific soft prompts is trained to disentangle knowledge acquisitions along with different dimensions and facilitate potential intra-dimension consolidation across CS-KG sources.
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- Boosting Variational Generative Model via Condition Enhancing and Lexical-Editing–
An Edit-CVAE (ECVAE) is proposed in which information-related data is exploited to address the problem by explicitly editing the generated sentence and enriching the latent representation while maintaining the diversity and information consistency.
- –
- –
- A NeuRetrieval Model for Human-Computer Conversations–
This paper proposes a novel retrieval model named NeuRetrieval for short text understanding, representation and semantic matching and investigates the effectiveness of the proposed deep neural network model for human-computer conversations.
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
Publication data from OpenAlex, with missing venues and authors filled in from Crossref; citation counts are the higher of OpenAlex and Semantic Scholar, last synced 2026-10-16. One-sentence summaries under some papers are written by Semantic Scholar’s model. Citation counts may be lower than on Google Scholar, which indexes more sources.
Report an error
Wrong papers, two people merged into one, or a profile that should not be here? Tell us and we will fix or hide it.