The rapid advancement of Large Language Models (LLMs) has driven significant progress in Natural Language Interface to Database (NLIDB). However, the widespread adoption of LLMs has raised critical privacy and security concerns. During interactions, LLMs may unintentionally expose confidential database contents or be manipulated by attackers to exfiltrate data through seemingly benign queries. While current efforts typically rely on rule-based heuristics or LLM agents to mitigate this leakage risk, these methods still struggle with complex inference-based attacks, suffer from high false positive rates, and often compromise the reliability of SQL queries. To address these challenges, we propose SafeNLIDB, a novel privacy-security alignment framework for LLM-based NLIDB. The framework features an automated pipeline that generates hybrid chain-of-thought interaction data from scratch, seamlessly combining explicit security reasoning with SQL generation. Additionally, we introduce reasoning warm-up and alternating preference optimization to overcome the multi-preference oscillations of Direct Preference Optimization (DPO), enabling LLMs to produce security-aware SQL through fine-grained reasoning without the need for human-annotated preference data. Extensive experiments demonstrate that our method outperforms both larger-scale LLMs and ideal-setting baselines, achieving significant security improvements while preserving high utility.

Continual Semantic Parsing (CSP) aims to train parsers to convert natural language questions into SQL across tasks with limited annotated examples, adapting to dynamically updated databases in real-world scenarios. Previous studies mitigate this challenge by replaying historical data or employing parameter-efficient tuning (PET), but they often violate data privacy or rely on ideal continual learning settings. To address these issues, we propose a new Large Language Model (LLM)-Enhanced Continuous Semantic Parsing method, named LECSP, which alleviates forgetting while encouraging generalization, without requiring real data replay or ideal settings. Specifically, it first analyzes the commonalities and differences between tasks from the SQL syntax perspective to guide LLMs in reconstructing key memories and improving memory accuracy through calibration. Then, it uses a task-aware dual-teacher distillation framework to promote the accumulation and transfer of knowledge during sequential training. Experimental results on two CSP benchmarks show that our method significantly outperforms existing methods, even those utilizing data replay or ideal settings. Additionally, we achieve generalization performance beyond upper limits, better adapting to unseen tasks.

Continual Semantic Parsing (CSP) enables parsers to generate SQL from natural language questions in task streams, using minimal annotated data to handle dynamically evolving databases in real-world scenarios. Previous works often rely on replaying historical data, which poses privacy concerns. Recently, replay-free continual learning methods based on Parameter-Efficient Tuning (PET) have gained widespread attention. However, they often rely on ideal settings and initial task data, sacrificing the model’s generalization ability, which limits their applicability in real-world scenarios. To address this, we propose a novel Adaptive PET eXpert meta-learning (APEX) approach for CSP. First, SQL syntax guides the LLM to assist experts in adaptively warming up, ensuring better model initialization. Then, a dynamically expanding expert pool stores knowledge and explores the relationship between experts and instances. Finally, a selection/fusion inference strategy based on sample historical visibility promotes expert collaboration. Experiments on two CSP benchmarks show that our method achieves superior performance without data replay or ideal settings, effectively handling cold start scenarios and generalizing to unseen tasks, even surpassing performance upper bounds.

Most existing document visual question answering (DocVQA) methods are restricted to single-page documents, limiting their applicability to more common multi-page scenarios. We introduce MP-FIRE, a multi-page DocVQA framework that integrates graph pruning-based reinforcement and cross-modal agent ensemble. MP-FIRE overcomes the Transformer’s inherent input length limitation, enabling the processing of an unlimited number of document pages in a single pass. Specifically, MP-FIRE employs a topology graph to extract document features and applies a two-stage pruning process to eliminate irrelevant document elements. It leverages the Performance Characterization Spectrum (PCS) to form a cross-modal agent ensemble, thereby enhancing complementary strengths and improving overall performance. Experimental results on DUDE and MP-DocVQA demonstrate MP-FIRE’s state-of-the-art performance, while ensuring strong generalization and fault tolerance.

Understanding the content of multi-page documents with rich layout information is a challenging task. Recent multimodal large language models (MLLMs) have made remarkable progress in understanding single-page document images. However, the understanding of multi-page documents remains insufficiently explored. This work proposes a Document Retrieval-enhanced, Expert-guided, Attention-aware Multimodal Framework, dubbed DREAM. Specifically, we propose a confidence-based, high-level semantic, multimodal retrieval method. Then, we propose a machine learning algorithm to complement the result of confidence-based retrieval and multimodal embedding similarity retrieval to obtain the most query-relevant set of document images. Subsequently, we designed a decoupled cross-page attention-aware multimodal language model for multi-page documents to interpret these retrieved images and produce the final answer. Experimental results demonstrate the effectiveness of the retrieval module within the framework, as well as the robust performance of the multimodal model in multi-page document comprehension. These findings offer a compelling solution for multi-page document comprehension and cross-page document visual question answering.

The rising popularity of mini-programs deployed on super-app platforms has drawn significant attention due to their convenience. However, developers' improper handling of data permission application in mini-programs has raised concerns about non-compliance and violations. Unfortunately, existing tools lack the capability to support the construction of a universal function call graph for the mini-program and the literature lacks a comprehensive and systematic study of the abusive issues. To bridge this gap, this paper introduces an automated tool, MiniChecker, to uncover the abusive permission request behavior in mini-programs. It defines five primary categories of abusive issues, namely homepage pop-up, overlaying pop-up, bothering pop-up, repeating pop-up, and looping pop-up, based on the request behavior features. MiniChecker achieves a detection precision rate of 82.4% and a recall rate of 95.3% on our benchmark, and identifies 3,866 risky mini-programs out of 20,000 real-world mini-programs. Our analysis reveals inherent design flaws in the mini-program permission mechanism, and we have shared our findings with several mini-program platforms.

Document Visual Question Answering (DVQA) involves responding to queries based on the contents of document images. Existing works are confined to locating information within a single page and lack support for cross-page question-and-answer interactions. Furthermore, the token length limitation on model inputs can lead to the truncation of answer-relevant segments. In this study, we present CREAM, an innovative methodology that focuses on high-performance retrieval and integrates relevant multimodal document information to effectively address this critical issue. To overcome the limitations of current text embedding similarity methods, we first employ a coarse-to-fine retrieval and ranking approach. The coarse phase calculates the similarity between the query and text chunk embeddings, while the fine phase involves multiple rounds of grouping and ordering with a large language model to identify the text chunks most relevant to the query. Subsequently, integrating an attention pooling mechanism for multi-page document images into the vision encoder allows us to effectively merge the visual information of multi-page documents, enabling the multimodal large language model (MLLM) to simultaneously process both single-page and multi-page documents. Finally, we apply various parameter-efficient tuning methods to enhance document visual question-answering performance. Experiments demonstrate that our approach secures state-of-the-art results across various document datasets.

Existing KG-augmented models for commonsense question answering primarily focus on designing elaborate Graph Neural Networks (GNNs) to model knowledge graphs (KGs).However, they ignore (i) the effectively fusing and reasoning over question context representations and the KG representations, and (ii) automatically selecting relevant nodes from the noisy KGs during reasoning.In this paper, we propose a novel model, JointLK, which solves the above limitations through the joint reasoning of LM and GNN and the dynamic KGs pruning mechanism.Specifically, JointLK performs joint reasoning between LM and GNN through a novel dense bidirectional attention module, in which each question token attends on KG nodes and each KG node attends on question tokens, and the two modal representations fuse and update mutually by multi-step interactions.Then, the dynamic pruning module uses the attention weights generated by joint reasoning to prune irrelevant KG nodes recursively.We evaluate JointLK on the Com-monsenseQA and OpenBookQA datasets, and demonstrate its improvements to the existing LM and LM+KG models, as well as its capability to perform interpretable reasoning 1 .

Table-based fact verification task aims to verify whether the given statement is supported by the given semi-structured table. Symbolic reasoning with logical operations plays a crucial role in this task. Existing methods leverage programs that contain rich logical information to enhance the verification process. However, due to the lack of fully supervised signals in the program generation process, spurious programs can be derived and employed, which leads to the inability of the model to catch helpful logical operations. To address the aforementioned problems, in this work, we formulate the table-based fact verification task as an evidence retrieval and reasoning framework, proposing the Logic-level Evidence Retrieval and Graph-based Verification network (LERGV). Specifically, we first retrieve logic-level program-like evidence from the given table and statement as supplementary evidence for the table. After that, we construct a logic-level graph to capture the logical relations between entities and functions in the retrieved evidence, and design a graph-based verification network to perform logic-level graph-based reasoning based on the constructed graph to classify the final entailment relation. Experimental results on the large-scale benchmark TABFACT show the effectiveness of the proposed approach.

Table-based fact verification is expected to perform both linguistic reasoning and symbolic reasoning.Existing methods lack attention to take advantage of the combination of linguistic information and symbolic information.In this work, we propose HeterTFV, a graph-based reasoning approach, that learns to combine linguistic information and symbolic information effectively.We first construct a program graph to encode programs, a kind of LISP-like logical form, to learn the semantic compositionality of the programs.Then we construct a heterogeneous graph to incorporate both linguistic information and symbolic information by introducing program nodes into the heterogeneous graph.Finally, we propose a graph-based reasoning approach to reason over the multiple types of nodes to make an effective combination of both types of information.Experimental results on a large-scale benchmark dataset TABFACT illustrate the effect of our approach.

Semantic information that has been proven to be necessary to the resolution of common noun phrases is typically ignored by most existing Chinese zero pronoun resolvers. This is because that zero pronouns convey no descriptive information, which makes it almost impossible to calculate semantic similarities between the zero pronoun and its candidate antecedents. Moreover, most of traditional approaches are based on the single-candidate model, which considers the candidate antecedents of a zero pronoun in isolation and thus overlooks their reciprocities. To address these problems, we first propose a neural-network-based zero pronoun resolver ( NZR ) that is capable of generating vector-space semantics of zero pronouns and candidate antecedents. On the basis of NZR , we develop the collaborative filtering-based framework for Chinese zero pronoun resolution task, exploring the reciprocities between the candidate antecedents of a zero pronoun to more rationally re-estimate their importance. Experimental results on the Chinese portion of the OntoNotes 5.0 corpus are encouraging: Our proposed model substantially surpasses the Chinese zero pronoun resolution baseline systems.

Natural Language Inference (NLI) is an active research area, where numerous approaches based on recurrent neural networks (RNNs), convolutional neural networks (CNNs), and self-attention networks (SANs) has been proposed. Although obtaining impressive performance, previous recurrent approaches are hard to train in parallel; convolutional models tend to cost more parameters, while self-attention networks are not good at capturing local dependency of texts. To address this problem, we introduce a Gaussian prior to selfattention mechanism, for better modeling the local structure of sentences. Then we propose an efficient RNN/CNN-free architecture named Gaussian Transformer for NLI, which consists of encoding blocks modeling both local and global dependency, high-order interaction blocks collecting the evidence of multi-step inference, and a lightweight comparison block saving lots of parameters. Experiments show that our model achieves new state-of-the-art performance on both SNLI and MultiNLI benchmarks with significantly fewer parameters and considerably less training time. Besides, evaluation using the Hard NLI datasets demonstrates that our approach is less affected by the undesirable annotation artifacts.

Deep neural network models for Chinese zero pronoun resolution learn semantic information for zero pronoun and candidate antecedents, but tend to be short-sightedthey often make local decisions.They typically predict coreference chains between the zero pronoun and one single candidate antecedent one link at a time, while overlooking their long-term influence on future decisions.Ideally, modeling useful information of preceding potential antecedents is critical when later predicting zero pronoun-candidate antecedent pairs.In this study, we show how to integrate local and global decision-making by exploiting deep reinforcement learning models.With the help of the reinforcement learning agent, our model learns the policy of selecting antecedents in a sequential manner, where useful information provided by earlier predicted antecedents could be utilized for making later coreference decisions.Experimental results on OntoNotes 5.0 dataset show that our technique surpasses the state-of-the-art models.

Recent neural network methods for zero pronoun resolution explore multiple models for generating representation vectors for zero pronouns and their candidate antecedents. Typically, contextual information is utilized to encode the zero pronouns since they are simply gaps that contain no actual content. To better utilize contexts of the zero pronouns, we here introduce the self-attention mechanism for encoding zero pronouns. With the help of the multiple hops of attention, our model is able to focus on some informative parts of the associated texts and therefore produces an efficient way of encoding the zero pronouns. In addition, an attention-based recurrent neural network is proposed for encoding candidate antecedents by their contents. Experiment results are encouraging: our proposed attention-based model gains the best performance on the Chinese portion of the OntoNotes corpus, substantially surpasses existing Chinese zero pronoun resolution baseline systems.

Abstract Neural Network-based approaches have recently produced good performances in Natural language tasks, such as Supertagging. In the supertagging task, a Supertag (Lexical category) is assigned to each word in an input sequence. Combinatory Categorial Grammar Supertagging is a more challenging problem than various sequence-tagging problems, such as part-of-speech (POS) tagging and named entity recognition due to the large number of the lexical categories. Specifically, simple Recurrent Neural Network (RNN) has shown to significantly outperform the previous state-of-the-art feed-forward neural networks. On the other hand, it is well known that Recurrent Networks fail to learn long dependencies. In this paper, we introduce a new neural network architecture based on backward and Bidirectional Long Short-Term Memory (BLSTM) Networks that has the ability to memorize information for long dependencies and benefit from both past and future information. State-of-the-art methods focus on previous information, whereas BLSTM has access to information in both previous and future directions. Our main findings are that bidirectional networks outperform unidirectional ones, and Long Short-Term Memory (LSTM) networks are more precise and successful than both unidirectional and bidirectional standard RNNs. Experiment results reveal the effectiveness of our proposed method on both in-domain and out-of-domain datasets. Experiments show improvements about (1.2 per cent) over standard RNN.

Existing approaches for Chinese zero pronoun resolution typically utilize only syntactical and lexical features while ignoring semantic information.The fundamental reason is that zero pronouns have no descriptive information, which brings difficulty in explicitly capturing their semantic similarities with antecedents.Meanwhile, representing zero pronouns is challenging since they are merely gaps that convey no actual content.In this paper, we address this issue by building a deep memory network that is capable of encoding zero pronouns into vector representations with information obtained from their contexts and potential antecedents.Consequently, our resolver takes advantage of semantic information by using these continuous distributed representations.Experiments on the OntoNotes 5.0 dataset show that the proposed memory network could substantially outperform the state-of-the-art systems in various experimental settings.

Existing approaches for Chinese zero pronoun resolution overlook semantic information. This is because zero pronouns have no descriptive information, which results in difficulty in explicitly capturing their semantic similarities with antecedents. Moreover, when dealing with candidate antecedents, traditional systems simply take advantage of the local information of a single candidate antecedent while failing to consider the underlying information provided by the other candidates from a global perspective. To address these weaknesses, we propose a novel zero pronoun-specific neural network, which is capable of representing zero pronouns by utilizing the contextual information at the semantic level. In addition, when dealing with candidate antecedents, a two-level candidate encoder is employed to explicitly capture both the local and global information of candidate antecedents. We conduct experiments on the Chinese portion of the OntoNotes 5.0 corpus. Experimental results show that our approach substantially outperforms the state-of-the-art method in various experimental settings.

In the age of Web 2.0, community user contributed questions and answers provide an important alternative for knowledge acquisition through web search. Question retrieval in current community-based question answering (CQA) services do not, in general, work well for long and complex queries, such as the questions. The main reasons are the verboseness in natural language queries and the word mismatch between the queries and the candidate questions in the CQA archive during retrieval. To address these two problems, existing solutions try to refine the search queries by distinguishing the key concepts in the queries and expanding the queries with relevant content. However, using the existing query refinement approaches can only identify the key and non-key concepts, while the differences between the key concepts are overlooked. Moreover, the existing query expansion approaches, not only overlook the weights of key concepts in the queries, but also fail to consider concept level expansion for them. In this paper, we explore a key concept identification approach for query refinement and a pivot language translation based approach to explore key concept paraphrasing. We further propose a new question retrieval model which can seamlessly integrate the key concepts and their paraphrases. The experimental results demonstrate that the integrated retrieval model significantly outperforms the state-of-the-art models in question retrieval.

Question retrieval in current community-based question answering (CQA) services does not, in general, work well for long and complex queries. One of the main difficulties lies in the word mismatch between queries and candidate questions. Existing solutions try to expand the queries at word level, but they usually fail to consider concept level enrichment. In this paper, we explore a pivot language translation based approach to derive the paraphrases of key concepts. We further propose a unified question retrieval model which integrates the keyconcepts and their paraphrases for the query question. Experimental results demonstrate that the paraphrase enhanced retrieval model significantly outperforms the state-of-the-art models in question retrieval.

In this paper, we propose a method to rank and assign weights to query terms according to their impact on the topic of the query. We use Search Result Overlap Ratio (SROR) to quantify the overlap of the search results of the full query and a shorten query after removing one term. Intuitively, if the overlap is small, it indicates a big topic shift and the removed term should be discriminative and important. The SROR could be used for measuring query term importance with a search engine automatically. By this way, learning based models could be trained based on a large number of automatically labeled instances and make predictions for future queries efficiently.