Spoken language understanding (SLU) is an essential part of a task-oriented dialogue system, which mainly includes intent detection and slot filling. Some existing approaches obtain enhanced semantic representation by establishing the correlation between two tasks. However, those methods show little improvement when applied to BERT, since BERT has learned rich features. In this paper, we propose BERT-based model with probability-aware gate mechanism, called PAGM ( <underline...

Multimodal aspect-based sentiment classification (MABSC) aims to identify the polarity toward specific aspects in multimodal data. It has gained significant attention with increasing use of social media platforms. Existing approaches primarily focus on analyzing content posts predict sentiment. However, they often struggle limited contextual information inherent posts, hindering accurate detection. To overcome this issue, we propose a novel dual cause analysis (MDCA) method track underlying...

This paper proposes a novel topic model, Author-Conference Topic-Connection (ACTC) Model for academic network search. The ACTC extends the author-conference-topic (ACT) model by adding subject of conference and latent mapping information between subjects topics. It simultaneously models topical aspects papers, authors conferences with two layers: layer corresponding to topic, word topic. Each author would be associated multinomial distribution over (eg., KM, DB, IR CIKM 2012),...

Recently, many Wikipedia-based methods have been proposed to improve the performance of different natural language processing (NLP) tasks, such as semantic relatedness computation, text classification and information retrieval. Among these methods, salient analysis (SSA) has proven be an effective way generate conceptual representation for words or documents. However, its feasibility effectiveness in retrieval is mostly unknown. In this paper, we study how efficiently use SSA performance,...

Pseudo‐relevance feedback is a well‐studied query expansion technique in which it assumed that the top‐ranked documents an initial set of retrieval results are relevant and terms then extracted from those documents. When selecting terms, most traditional models do not simultaneously consider term frequency co‐occurrence relationships between candidate terms. Intuitively, however, has higher with more likely to be related topic. In this article, we propose kernel co‐occurrence‐based framework...

At present, spoken language understanding (SLU) in multi-turn dialogue is a research hotspot, which mainly includes intent detection and slot filling. SLU models trained by large-scale corpus can learn good superficial semantic grammatical information. But they lack the ability for modeling knowledge needed to understand language. In order further deep information of dialogue, external needs be modeled incorporated into model. addition, utilizing correlation between history current utterance...

This paper presents a tag-topic model with Dirichlet Forest prior (TTM-DF) for semantic knowledge acquisition from blog. The TTM-DF extends the (TTM) by replacing over topic-word multinomial. correlation between words are calculated to generate set of Must-Links and Cannot-Links, then structures trees obtained though encoding constraints Cannot-Links. Words under same subtrees expected be more correlated than different subtrees. We conduct experiments on synthetic blog dataset. Both...

Most of the existing information retrieval models assume that terms a text document are independent each other. These integrate three major variables to determine degree importance term for document: within frequency, length and specificity in collection. Intuitively, is not only dependent on aspects mentioned above, but also semantic coherence between document. In this paper, we propose heuristic approach, which query with adopted improve performance. Experimental results standard TREC...

Dialogue state tracking (DST) is a core component of task-oriented dialogue systems. Recent works focus mainly on end-to-end DST models that omit the spoken language understanding (SLU) module to directly obtain based user's dialogue. However, slot information detected by filling in SLU closely tied slot-value pair needs be updated DST. Efficient use key semantic knowledge obtained contributes improving performance Based this idea, we introduce as subtask and build an joint model explicitly...

Recent years have witnessed a substantial increase in the use of deep learning to solve various natural language processing (NLP) problems. Early models were constrained by their sequential or unidirectional nature, such that they struggled capture contextual relationships across text inputs. The introduction bidirectional encoder representations from transformers (BERT) leads robust for transformer model can understand broader context and deliver state-of-the-art performance NLP tasks. This...

The scaled Web 3.0 digital economy, represented by decentralized finance (DeFi), has sparked increasing interest in the past few years, which usually relies on blockchain for token transfer and diverse transaction logic. However, illegal behaviors, such as financial fraud, hacker attacks, money laundering, are rampant ecosystem seriously threaten its integrity security. In this paper, we propose a novel double graph-based Ethereum account de-anonymization inference method, dubbed DBG4ETH,...

Knowledge Graph Completion (KGC) has garnered massive research interest recently, and most existing methods are designed following a transductive setting where all entities observed during training. Despite the great progress on KGC, these struggle to conduct reasoning emerging KGs involving unseen entities. Thus, inductive which aims deduce missing links among entities, become new trend. Many studies transform KGC as graph classification problem by extracting enclosing subgraphs surrounding...