基于ChatGPT的电信诈骗案件类型影响力评估

doi:10.3778/j.issn.1673-9418.2306044

摘要/Abstract

摘要： 当前电信诈骗犯罪呈高发态势，严重威胁人民群众财产安全，为了优化反诈策略、客观准确分析不同类型电信诈骗案件的发案趋势和发案特征，确定影响力较大的犯罪手段和方式，提出一种基于ChatGPT的电信诈骗案件类型影响力评估方法。借助知识图谱使案件文本内容结构化，并将案发时间、涉案金额、涉案事主人数作为评估案件影响力的因素，对电信诈骗方式进行量化。首先利用ChatGPT通过多轮问答的形式对电信诈骗案件文本语料进行数据预处理和知识抽取，低资源、快速及时地构建电信诈骗领域的案件知识图谱，并基于知识图谱统计分析案发时间、涉案金额、涉案事主人数等各类因素，把不同案件类型的影响抽象为影响因子，用影响因子刻画案发趋势与发案特征，以进行综合分析研判。通过对现有案例数据进行分析，计算案件类型影响因子，得到了不同案件类型的影响因子变化，验证了影响因子计算方法的科学性与有效性，为电信诈骗类型的评估提供了一种新的方法和思路。结合ChatGPT与知识图谱的优势，有助于及时把握案件发展变化趋势，为打击电信诈骗提供有力的支持和指导，对于保护公众的财产安全和社会的稳定具有重要意义。

关键词: 电信诈骗, ChatGPT, 知识图谱, 影响评估

Abstract: At present, telecommunications fraud crimes are on the rise, posing a serious threat to the safety of people??s property. In order to optimize anti-fraud strategies, objectively and accurately analyze the trends and characteristics of different types of telecommunications fraud cases, and determine the most influential criminal methods, a ChatGPT based telecommunications fraud case type impact assessment method is proposed. By utilizing a knowledge graph, the content of the case text is structured, and the methods of telecommunications fraud are quantified by taking the time of the incident, the amount involved, and the number of individuals involved as factors to evaluate the impact of the case. Firstly, ChatGPT is used to preprocess and extract knowledge from the text corpus of telecommu-nications fraud cases through multiple rounds of Q&A, in order to quickly and timely construct a case knowledge graph in the field of telecommunications fraud with low resources. Based on the knowledge graph, various factors such as incident time, amount involved, and the number of involved parties are statistically analyzed, and the impact of different types of cases is abstracted into influencing factors. The influencing factors are used to depict the trend and characteristics of incidents, to conduct comprehensive analysis and judgment. This paper analyzes existing case data and calculates the impact factors of case types, obtaining the changes in impact factors of different case types, verifying the scientific and effective calculation methods of impact factors, and providing a new method for the evaluation of telecommunications fraud types. Combining the advantages of ChatGPT and knowledge graph helps to timely grasp the trend of case development and changes, provides strong support and guidance to combat teleco-mmunications fraud, and is of great significance for protecting public property safety and social stability.

Key words: Telecom fraud, ChatGPT, knowledge graph, impact assessment

裴炳森, 李欣, 吴越. 基于ChatGPT的电信诈骗案件类型影响力评估[J]. 计算机科学与探索, 2023, 17(10): 2413-2425.

PEI Bingsen, LI Xin, WU Yue. Influence Evaluation of Telecom Fraud Case Types Based on ChatGPT[J]. Journal of Frontiers of Computer Science and Technology, 2023, 17(10): 2413-2425.

参考文献

[1] BROWN T B, MANN B, RYDER N, et al. Language models are few-shot learners[C]//Advances in Neural Information Processing Systems 33, Dec 6-12, 2020: 1877-1901.
[2] 卢经纬, 郭超, 戴星原, 等. 问答ChatGPT之后:超大预训练模型的机遇和挑战[J]. 自动化学报, 2023, 49(4): 705-717.
LU J W, GUO C, DAI X Y, et al. The ChatGPT after: opportunities and challenges of very large scale pre-trained models[J]. Acta Automatica Sinica, 2023, 49(4): 705-717.
[3] 桑基韬, 于剑. 从ChatGPT看AI未来趋势和挑战[J]. 计算机研究与发展, 2023, 60(6): 1191-1201.
SANG J T, YU J. ChatGPT: a Glimpse into AI’s future[J]. Journal of Computer Research and Development, 2023, 60(6): 1191-1201.
[4] WU C, YIN S, QI W, et al. Visual ChatGPT: talking, drawing and editing with visual foundation models[J]. arXiv:2303.04671, 2023.
[5] YANG Z, LI L, WANG J, et al. MM-ReAct: prompting ChatGPT for multimodal reasoning and action[J]. arXiv:2303.11381, 2023.
[6] SHEN Y, SONG K, TAN X, et al. HuggingGPT: solving AI tasks with ChatGPT and its friends in huggingface[J]. arXiv:2303.17580, 2023.
[7] TOUVRON H, LAVRIL T, IZACARD G, et al. LLaMA: open and efficient foundation language models[J]. arXiv:2302.13971, 2023.
[8] DU Z, QIAN Y, LIU X, et al. GLM: general language model pretraining with autoregressive blank infilling[J]. arXiv:2103.10360, 2021.
[9] ZHU P, CHENG D, YANG F, et al. Improving Chinese named entity recognition by large-scale syntactic dependency graph[J]. IEEE/ACM Transactions on Audio, Speech, and Language Processing, 2022, 30: 979-991.
[10] ZHONG Q, TANG Y. Chinese named entity recognition based on gated graph neural network[C]//Proceedings of the 14th International Conference on Knowledge Science, Engineering and Management, Tokyo, Aug 14-16, 2021: 604-613.
[11] 宋旭晖, 于洪涛, 李邵梅. 基于图注意力网络字词融合的中文命名实体识别[J]. 计算机工程, 2022, 48(10): 298-305.
SONG X H，YU H T，LI S M. Chinese named entity recog-nition based on word fusion of graph attention network[J]. Computer Engineering, 2022, 48(10): 298-305.
[12] 段建勇, 朱奕霏, 王昊, 等. 基于位置嵌入多级预测的中文嵌套命名实体识别[J/OL]. 计算机工程 [2023-05-06]. DOI:10.19678/j.issn.1000-3428.0066379.
DUAN J Y, ZHU Y F, WANG H, et al. Chinese nested named entity recognition based on location-embedded multilevel prediction[J/OL]. Computer Engineering [2023-05-06]. DOI:10.19678/j.issn.1000-3428.0066379.
[13] WANG D, TIWARI P, GARG S, et al. Structural block driven enhanced convolutional neural representation for relation extraction[J]. Applied Soft Computing, 2020, 86: 105913.
[14] 郝小芳, 张超群, 李晓翔, 等. 融合交互注意力网络的实体和关系联合抽取模型[J/OL]. 计算机工程与应用 [2023-05-08]. http://kns.cnki.net/kcms/dtail/11.2127.TP.20230506.1624. 020.html.
HAO X F, ZHANG C Q, LI X X, et al. Joint entity relation extraction model based on interactive attention[J/OL]. Com-puter Engineering and Applications [2023-05-08]. http://kns.cnki.net/kcms/dtail/11.2127.TP.20230506.1624.020.html.
[15] 许亮, 张春, 张宁, 等. 融合多Prompt模板的零样本关系抽取模型[J/OL]. 计算机应用[2023-05-08]. http://kns.cnki.net/kcms/detail/51.1307.TP.20230504.1044.018.html.
XU L, ZHANG C, ZHANG N, et al. Zero-shot relation ext-raction model via multi-template fusion in Prompt[J/OL]. Journal of Computer Applications [2023-05-08]. http://kns.cnki.net/kcms/detail/51.1307.TP.20230504.1044.018.html.
[16] LI X, YIN F, SUN Z, et al. Entity-relation extraction as multi-turn question answering[J]. arXiv:1905.05529, 2019.
[17] HUANG Z, XU W, YU K. Bidirectional LSTM-CRF models for sequence tagging[J]. arXiv:1508.01991, 2015.
[18] SOUZA F, NOGUEIRA R, LOTUFO R. Portuguese named entity recognition using BERT-CRF[J]. arXiv:1909.10649, 2019.
[19] XUAN Z, BAO R, JIANG S. FGN: fusion Glyph network for Chinese named entity recognition[C]//Proceedings of the 5th China Conference on Knowledge Graph and Semantic Computing: Knowledge Graph and Cognitive Intelligence, Nanchang, Nov 12-15, 2020: 28-40.
[20] ZHOU Y, ZHENG X, HUANG X. Chinese named entity recog-nition augmented with lexicon memory[J]. arXiv:1912.08282, 2019.
[21] WU S, SONG X, FENG Z. MECT: multi-metadata embed-ding based cross-transformer for Chinese named entity recog-nition[J]. arXiv:2107.05418, 2021.
[22] REN X, ZHOU P, MENG X, et al. PanGu-??: towards trillion parameter language model with sparse heterogeneous com-puting[J]. arXiv:2303.10845, 2023.
[23] FU T J, LI P H, MA W Y. GraphRel: modeling text as rela-tional graphs for joint entity and relation extraction[C]//Pro-ceedings of the 57th Annual Meeting of the Association for Computational Linguistics, Florence, Jul 28-Aug 2, 2019: 1409-1418.
[24] XIANG R Z, SHI Z H, DAOJIAN Z, et al. Learning the extraction order of multiple relational facts in a sentence with reinforcement learning[C]//Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing, Hong Kong, China, Nov 3-7, 2019: 367-377.
[25] WEI Z P. A novel cascade binary tagging framework for relational triple extraction[C]//Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, Jul 5-10, 2020:?1476-1488.