Abstract:This paper designs a dialogue model based on the granulated knowledge graph (Granulation Knowledge aware Dialogue Model, GKDM). Through the method of structural granulation, the knowledge graph is divided into two modules: local knowledge graph and non-local knowledge graph, respectively capturing the local semantics adjacent to the knowledge entity and the overall semantic representation of the located knowledge sub-graph, which solves the problem of the division of the knowledge graph. This paper designs the static graph attention mechanism and the hierarchical dynamic graph attention mechanism to participate in the process of encoding and decoding in the model respectively, and aggregates the semantics of knowledge entities through multi-level attention weights, and fuses them into the text vector, which solves the problem of the fusion of multi-layer knowledge information. Based on the above two innovations, the model proposed in this paper alleviates the problems of the current knowledge graph-based dialogue model that cannot capture the reasoning meaning of the multi-hop paths of the knowledge graph and the inconsistent theme of the reply context.

Alert Name:
Alerting to:
Authentication email will be sent to your email address in 24 hours
Frequency:
Email Message Format:	Plain text Graphical(HTML)

Complete the form below and we will recommend the selected titles to your friends on your behalf. * Indicates a required field.
Your name*:
Your email address*:
Recipient's name*:
Recipient's email address*:
(multiple recipient's names and email addresses should be separated with semicolons)
Your comments:	I thought you would find the page(s) useful.

Your name:
Your email address:
Recipient's name:
Recipient's email address:
(multiple recipient's names and email addresses should be separated with semicolons)
Your comments:	I thought you would find this page useful.

Disclaimer: This message was sent to your friend using the "Send it to a friend" facility on the Sciencepaper Online’ WWW site, http://www.paper.edu.cn/en. The Sciencepaper Online is not responsible for the content of this email, and anything said in this email does not necessarily reflect the Sciencepaper Online's views.


	1. Granularized Dialogue Model of The Knowledge Graph
	Zhao Meiyong,Yu Wen
	Computer Science and Technology 10 April 2024
	Show/Hide Abstract \| Cite this paper︱Full-text: PDF (0 B)

	Abstract:This paper designs a dialogue model based on the granulated knowledge graph (Granulation Knowledge aware Dialogue Model, GKDM). Through the method of structural granulation, the knowledge graph is divided into two modules: local knowledge graph and non-local knowledge graph, respectively capturing the local semantics adjacent to the knowledge entity and the overall semantic representation of the located knowledge sub-graph, which solves the problem of the division of the knowledge graph. This paper designs the static graph attention mechanism and the hierarchical dynamic graph attention mechanism to participate in the process of encoding and decoding in the model respectively, and aggregates the semantics of knowledge entities through multi-level attention weights, and fuses them into the text vector, which solves the problem of the fusion of multi-layer knowledge information. Based on the above two innovations, the model proposed in this paper alleviates the problems of the current knowledge graph-based dialogue model that cannot capture the reasoning meaning of the multi-hop paths of the knowledge graph and the inconsistent theme of the reply context.
	TO cite this article:Zhao Meiyong,Yu Wen. Granularized Dialogue Model of The Knowledge Graph[OL].[10 April 2024] http://en.paper.edu.cn/en_releasepaper/content/4763289


	2. A Multi-Document Inference Method Based on TR-BERT and Attention Networks
	ZHAO Jiaqi,LIN Rongheng
	Computer Science and Technology 13 March 2024
	Show/Hide Abstract \| Cite this paper︱Full-text: PDF (0 B)

	Abstract:In the task of multi-document inference, information related to the answer may reside across multiple relevant texts, and sometimes this information is not directly associated with the question. To address the challenge of balancing accuracy and efficiency in multi-document inference for the electric utility customer service scenario, this paper proposes a multi-level inference method based on pre-trained models and attention networks. The model utilizes pre-trained models to preserve the rich semantics extracted from paragraph texts and questions, and then evaluates the relevance of candidates through an attention mechanism. Furthermore, due to the real-time requirements of the question-answering scenario, we employ the TR-BERT pre-trained model based on dynamic token reduction and simplify the attention network. Experimental results on the WikiHop dataset demonstrate that the model overall exhibits advantages in both computational speed and accuracy, providing effective methodological support for the multi-turn question-answering functionality in intelligent question-answering systems.
	TO cite this article:ZHAO Jiaqi,LIN Rongheng. A Multi-Document Inference Method Based on TR-BERT and Attention Networks[OL].[13 March 2024] http://en.paper.edu.cn/en_releasepaper/content/4762558

Saved Papers

Saved Papers


	3. Named Entity Recognition in the Perovskite Field Based on Convolutional Neural Networks and MatBERT
	ZHANG Jiaxin,ZHANG Lingxue,SUN Yuxuan,LI Wei,QUHE Ruge
	Computer Science and Technology 13 March 2024
	Show/Hide Abstract \| Cite this paper︱Full-text: PDF (0 B)

	Abstract:Due to the significant increase in publications in the field of materials science, there has been a bottleneck in organizing material science knowledge and discovering new materials. The number of literature in the emerging field of perovskite materials has grown to a massive scale. It is necessary to compile information on the structure, properties, synthesis methods, characterization techniques, and applications of perovskite materials. To address this issue, we employ named entity recognition, a natural language processing technique, to extract important entities from perovskite material texts. In this paper, we propose a method based on convolutional neural networks (CNN) and MatBERT. Firstly, we utilize MatBERT, which has been pre-trained on a large amount of material science text, to generate contextualized word embeddings. Next, we extract feature information using a CNN model. Finally, a conditional random field (CRF) layer is used for decoding sequences in addition to calculating the training and validation loss. Experimental results demonstrate that the performance of our model on perovskite material dataset is improved by 1%~6% compared with BERT, SciBERT and MatBERT models. Through this model, we extract the entities of 2389 abstracts to obtain knowledge of perovskite materials.
	TO cite this article:ZHANG Jiaxin,ZHANG Lingxue,SUN Yuxuan, et al. Named Entity Recognition in the Perovskite Field Based on Convolutional Neural Networks and MatBERT[OL].[13 March 2024] http://en.paper.edu.cn/en_releasepaper/content/4762696


	4. Enhancing Persona Consistency in Dialogue Generation Algorithm with Retrieval Augmentation
	SHI Haozhe
	Computer Science and Technology 12 March 2024
	Show/Hide Abstract \| Cite this paper︱Full-text: PDF (0 B)

	Abstract:Open-domain dialogue systems are designed to fulfill people\'s daily communicative and emotional requirements, with the goal of cultivating long-term relationships with users. Yet, these systems encounter challenges in sustaining persona consistency, as responses generated at times are not logically aligned with the established character persona or preceding dialogues. This discrepancy undermines dialogue coherence and emotional engagement, consequently impeding the development of profound connections with users. Addressing this issue, this study introduce an innovative dialogue generation algorithm that incorporates retrieval-augmentation techniques. By forming an database of character information, the algorithm aids large language models in retrieving persona-relevant data during interactions, ensuring that responses consistently align with the character\'s defined persona. This method significantly mitigates the occurrence of generating inconsistent response, an "hallucination" effect. This study demonstrates the substantial impact of information optimization and filtering mechanisms on enhancing persona consistency within dialogue systems, as evidenced through comprehensive evaluation across three pivotal performance metrics: information relevance, faithfulness, and reponse relevance, facilitated by an integration of various retrieval strategies and information optimization techniques.
	TO cite this article:SHI Haozhe. Enhancing Persona Consistency in Dialogue Generation Algorithm with Retrieval Augmentation[OL].[12 March 2024] http://en.paper.edu.cn/en_releasepaper/content/4762669


	5. CPBNet：Concentrate,Parallel and Bimodal Network for Logistics Scene Text Detection and Recognition
	MA Yu-Chen
	Computer Science and Technology 28 February 2024
	Show/Hide Abstract \| Cite this paper︱Full-text: PDF (0 B)

	Abstract:In the logistics industry, express sorting has always been an important link in ensuring smooth transportation of logistics. The recognition of logistics forms directly determines the sorting efficiency, but how to effectively improve the accuracy of text recognition of logistics forms in complex sorting environments is still a research challenge. In this article, we believe that the limitations of existing text models in logistics sorting scenarios mainly come from: 1) the interference caused by complex environments; 2) the incompatibility of recognition accuracy and speed; 3) single-modal limitation. Accordingly, this paper proposes CPBnet based on the principles of concentrate,parallel and bimodal. First, we corrected the form angularly, geometrically, and photometrically for complex scenes. Then, using a parallel method, the Attention mechanism is added to the visual model to guide the CTC training strategy, and the more accurate characteristics of the Attention model are used to train the backbone network to obtain better convolution features, and then the CTC branch is used for prediction, thereby ensuring Speed at inference. Finally, a language model is added after the visual model for semantic correction. The language model fully learns the input contextual information to make up for the visual semantic deficiency. There are basically very few pictures of sorting scenes in existing general text data sets. The lack of data in the field of sorting scenes has created a bottleneck for the application of deep learning in sorting scenes. Therefore, this article simulates real form data, prepares a sorting scene data set by itself, and proves through a large number of experiments that CPBNet has advantages on this data set and achieves the most advanced results.
	TO cite this article:MA Yu-Chen. CPBNet：Concentrate,Parallel and Bimodal Network for Logistics Scene Text Detection and Recognition[OL].[28 February 2024] http://en.paper.edu.cn/en_releasepaper/content/4762301


	6. Relation Extraction Method for Chinese Public Opinion Based on Transferring Pre-trained Models and Merging Multiple Features
	ZHANG Yunkai,CHENG Bo
	Computer Science and Technology 27 February 2024
	Show/Hide Abstract \| Cite this paper︱Full-text: PDF (0 B)

	Abstract:The public opinion has profoundly and extensively impacted the society, making the research on relation extraction(RE) in the field of public opinion crucial. Many existing relation extraction models either focus solely on basic information of Chinese characters or fail to fully leverage pre-trained models for extraction. Therefore, this paper proposes a relation extraction model, CwTransRE, which incorporates basic Chinese character information, glyph information, pinyin information and Chinese word information through transferring pre-trained models. CwTransRE enhances the effectiveness of relation extraction in two key aspects: firstly, besides basic Chinese character information, the integrated glyph, pinyin and Chinese word information enriches the semantic features of embeddings; secondly, the introduction of pre-trained models aids in obtaining more accurate embeddings, especially when dealing with relatively small training datasets. Experimental results on an open-source public opinion dataset demonstrate that our model achieves an F1 score of 0.703, outperforming NovelTagging, GraphRel(1p), GraphRel(2p) , TAG-JE and CasRel.
	TO cite this article:ZHANG Yunkai,CHENG Bo. Relation Extraction Method for Chinese Public Opinion Based on Transferring Pre-trained Models and Merging Multiple Features[OL].[27 February 2024] http://en.paper.edu.cn/en_releasepaper/content/4762184


	7. Data Augmentation Based Biaffine Attention Coreference Resolution Algorithm
	LI Yuanxin
	Computer Science and Technology 11 May 2023
	Show/Hide Abstract \| Cite this paper︱Full-text: PDF (0 B)

	Abstract:Event text generally has a variety of descriptions. Event coreference resolution aims to find the similarity matching between two kinds of event text descriptions, which can help professionals find the same event from multi-source information. We propose a data augmentation based biaffine attention Coreference resolution algorithm, DEBACR. Roberta was selected as the pre-training model for feature extraction. Various data enhancement methods such as back translation and FGM were used to enhance the generalization ability of the model. We propose a fusion model of global and local information, which combines the coreference relation judgment of trigger word consistency, event elements consistency and event overall consistency. By using the biaffine attention mechanism, the important features are screened out for resolution and classification. Experimental results show that the accuracy of the algorithm is effectively improved in multiple Chinese event coreference resolution datasets than semantic similarity model based on Siamese network.
	TO cite this article:LI Yuanxin. Data Augmentation Based Biaffine Attention Coreference Resolution Algorithm[OL].[11 May 2023] http://en.paper.edu.cn/en_releasepaper/content/4760702


	8. A Graph Learning based Approach for Similar Text Recommendation in Enforcement Document of Urban Management
	Liu Huiyong,Xiong Songlin
	Computer Science and Technology 31 March 2023
	Show/Hide Abstract \| Cite this paper︱Full-text: PDF (0 B)

	Abstract:The research on text classification is the most basic part of the natural language processing sector in the field of machine learning. Since the topic was proposed, quite a few methods have been produced to solve this problem. However, most of the current text classification methods are for text classification of short texts or long texts containing a certain specific content. There are relatively few studies on text classification methods that contain multiple texts and specific structural structures. In the professional field of urban management and law enforcement documents, the composition of documents has the characteristics of multiple texts and specific structural structures. In order to complete the text classification work of this type of task, it is necessary to adopt some targeted methods. In the past decade, deep learning has achieved great success in various fields. In order to improve the accuracy and efficiency of such text classification, methods of introducing graph structure and deep learning are considered. Combining the characteristics of such texts, this paper mainly studies the graph convolutional neural network algorithm for such text classification. In order to extract more information from the text, construct a graph on the entire corpus, use the vocabulary in the text and the text itself as nodes in the graph, use the vocabulary co-occurrence matrix information to construct the edge between texts, and use the vocabulary The edge between words and words is constructed by using the position context relationship between them, and the edge between words and text is constructed by using the frequency of words appearing in the text, and then the multi-text classification problem is regarded as a graph node classification problem. After iteration and optimization of the algorithm, the final conclusion is that the method of using the graph convolutional neural network can greatly improve the accuracy of this type of multi-text classification within the acceptable efficiency loss range. And further improve the accuracy of text recommendation and tailoring suggestions in the follow-up work.
	TO cite this article:Liu Huiyong,Xiong Songlin. A Graph Learning based Approach for Similar Text Recommendation in Enforcement Document of Urban Management[OL].[31 March 2023] http://en.paper.edu.cn/en_releasepaper/content/4759931


	9. Using Pre-Trained Language Models for Aspect Category Sentiment Analysis in a Simple Causal Inference way
	XIAO Zhiyi
	Computer Science and Technology 22 March 2023
	Show/Hide Abstract \| Cite this paper︱Full-text: PDF (0 B)

	Abstract:Aspect category sentiment analysis (ACSA) is a fine-grained sentiment analysis task that aims to identify the sentiment polarity to wards an explicit or implicit aspect (term) in a sentence. While pre trained language models like BERT have shown promising results in this area, they can also have inaccuracies caused by spurious correlations that do not necessarily reflect the sentiment. The bias in pre-trained language models and the unbalanced distribution of the finetuning data always leads to high accuracy in test data, but it could perform poorly on the prediction in reality. Inspired by the stability of causality, we propose counterfactual ACSA (CF-ACSA), a model-agnostic framework that reduces language bias by separating the direct language effect from the total effect on sentiment. We first identify the spurious correlations be tween term and sentiment polarity through a semi-supervised approach. Next, we use the spurious term we find to construct counterfactual sam ples for the original data. Finally, we make causal inferences and get an unbiased result between counterfactual samples and the original data. Extensive experiments demonstrate the effctiveness of our model, and results show that it outperforms the state-of-the-art methods on ASAP datasets.
	TO cite this article:XIAO Zhiyi. Using Pre-Trained Language Models for Aspect Category Sentiment Analysis in a Simple Causal Inference way[OL].[22 March 2023] http://en.paper.edu.cn/en_releasepaper/content/4759903


	10. Stroke-encoding and Pinyin-learning Enhanced Chinese Pre-trained Language Representation Model
	ZHOU Tong
	Computer Science and Technology 22 March 2023
	Show/Hide Abstract \| Cite this paper︱Full-text: PDF (0 B)

	Abstract:Language models pre-trained on large unlabeled corpora have proven to be very effective in improving many downstream NLP tasks. However, existing language models are primarily designed for English, and less consideration has been given to the more abundant semantic information that Chinese characters imply. The unique semantically related stroke sequence pattern and polyphony enable the enhancement of a Chinese language representation model. Masked language models, such as BERT, are also plagued by inefficient training data utilization, requiring more iterations to complete training. In light of these shortcomings, we propose an improved, customized Chinese pre-trained language model based on the transformer, called SPCLM (Stroke-encoding and Pinyin-learning enhanced Chinese pre-trained Language representation Model). SPCLM contains stroke encoders and an additional pronunciation prediction task. Moreover, the autoregressive objection and mask prediction jointly assist in model formulation. Experimental results demonstrate that SPCLM outperforms other baseline methods, achieving comparable results on five Chinese NLP tasks, with insufficient pre-training, including natural language inference, semantic similarity, named entity recognition, sentiment analysis, and question answering.
	TO cite this article:ZHOU Tong. Stroke-encoding and Pinyin-learning Enhanced Chinese Pre-trained Language Representation Model[OL].[22 March 2023] http://en.paper.edu.cn/en_releasepaper/content/4759900

	Check out RSS, or use RSS reader to subscribe this item