Research on knowledge point relationship extraction for elementary mathematics
-
摘要: 随着互联网技术的发展,在线教育已经改变了学生的学习方式.但由于缺乏完整的知识体系,在线教育存在着智能化程度低和“信息迷航”的问题.因此,构建知识体系成为在线教育平台的核心技术.知识点间的关系提取是知识体系构建的主要任务之一,目前比较高效的关系提取算法主要是监督式的.但是这类方法受限于文本质量低、语料稀缺、标签数据难获取、特征工程效率低、难以提取有向关系等挑战.为此,基于百科语料和远程监督思想,研究了知识点间的关系提取算法.提出了基于关系表示的注意力机制,该方法能够提取知识点间的有向关系信息.结合了GCN和LSTM的优势,提出了GCLSTM,该模型更好地提取了句子中的多点信息.基于Transformer架构和关系表示的注意力机制,提出了适用于有向关系提取的BTRE模型,降低了模型的复杂度.设计并实现了知识点关系提取系统.通过设计3组对比实验,验证了模型的性能和效率.
-
关键词:
- 知识体系构建 /
- 关系提取 /
- 注意力机制 /
- 远程监督 /
- Transformer
Abstract: With the development of Internet technology, online education has changed the learning style of students. However, given the lack of a complete knowledge system, online education has a low degree of intelligence and a/knowledge trek0problem. The relation-extraction concept is one of the key elements of knowledge system construction. Therefore, building knowledge systems has become the core technology of online education platforms. At present, the more efficient relationship extraction algorithms are usually supervised. However, such methods suffer from low text quality, scarcity of corpus, difficulty in labeling data, low efficiency of feature engineering, and difficulty in extracting directional relationships. Therefore, this paper studies the relation-extraction algorithm between concepts based on an encyclopedic corpus and distant supervision methods. An attention mechanism based on relational representation is proposed, which can extract the forward relationship information between knowledge points. Combining the advantages of GCN and LSTM, GCLSTM is proposed, which better extracts multipoint information in sentences. Based on the attention mechanism of Transform architecture and relational representation, a BTRE model suitable for the extraction of directional relationships is proposed, which reduces the complexity of the model. Hence, a knowledge point relationship extraction system is designed and implemented. The performance and efficiency of the model are verified by designing three sets of comparative experiments. -
表 1 关系样例
Tab. 1 Relationship example
共现句$s$ 知识点$e_{1}$ 知识点$e_{2}$ 关系$r$ 两个整数的最大公因子可用于计算两数的最小公倍数 最小公倍数 整数 依赖 求几个整数的最大公因数, 只要把它们所有共有的质因数连乘 质因数 最大公因数 被依赖 代数式根据它所包含的运算可以分为有理式和无理式, 术语 而有理式又可以分为整式和分式 整式 有理式 正比例函数为特殊的一次函数 一次函数 正比例函数 包含 实数根也经常被称为实根 实数根 实根 同义 分数分为两类:真分数和假分数 真分数 假分数 反义 直角三角形的外心在三角形斜边中点 直角三角形 外心 拥有 二元二次方程是含有两个未知数且未知数的最高次数为2的整式方程 未知数 二元二次方程 被拥有 表 2 自注意力机制、循环结构、卷积操作复杂度对比
Tab. 2 Comparison of self-attention mechanism, circulation mechanism, and convolution operation complexity
运算方式 计算复杂度 串行化操作数 最大路径长度 自注意力机制 $O(l\cdot d^{x})$ $O($1) $O($1) 循环结构 $O(l\cdot (d^{x})^{2})$ $O(l)$ $O(l)$ 卷积操作 $O(k\cdot l\cdot (d^{x})^{2})$ $O($1) $O($log$_{k}(l))$ 表 3 BTRE模型与GCLSTM模型、PCNN模型实验结果对比
Tab. 3 Comparison of BTRE model with GCLSTM model and PCNN model
模型名称 准确率 召回率 F1 AUC PCNN 74.15% 65.83% 69.74% 0.548 GCLSTM 74.73% 69.33% 71.93% 0.569 BTRE 75.35% 70.07% 72.61% 0.623 表 4 BTRE模型中不同子层数量结果对比
Tab. 4 Comparison of the number of different sub-layers in the BTRE model
子层数量N 准确率 召回率 F1 AUC 训练集F1 1 68.78% 64.84% 66.75% 0.543 91.35% 2 75.35% 70.07% 72.61% 0.624 93.26% 3 71.05% 67.33% 69.14% 0.602 95.33% 4 69.90% 68.33% 69.10% 0.573 95.78% -
[1] LIU H, MA W, YANG Y, et al. Learning concept graphs from online educational data[J]. Journal of Artificial Intelligence Research, 2016, 55:1059-1090. doi: 10.1613/jair.5002 [2] NOVAK J D, BOB GOWIN D, JOHANSEN G T. The use of concept mapping and knowledge vee mapping with junior high school science students[J]. Science education, 1983, 67(5):625-645. doi: 10.1002/sce.3730670511 [3] MIWA M, BANSAL M. End-to-End Relation Extraction using LSTMs on Sequencesand Tree Structures[C]//Proceedings of the 54th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), 2016: 1105-1116. [4] MINTZ M, BILIS S, SNOW R, et al. Distant supervision for relation extraction without labeled data[C]//Proceedings of the Joint Conference of the 47th Annual Meeting of the ACL and the 4th International Joint Conference on Natural Language Processing of the AFNLP: Volume 2, Association for Computational Linguistics, 2009: 1003-1011. [5] ZHOU P, SHI W, TIAN J, et al. Attention-based bidirectional long short-term memory networks for relation classification[C]//Proceedings of the 54th Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers), 2016: 207-212. [6] LIN Y, SHEN S, LIU Z, et al. Neural relation extraction with selective attention over instances[C]//Proceedings of the 54th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), 2016: 2124-2133. [7] LUONG M T, PHAM H, MANNING C D. Effective approaches to attention-based neural machine translation[C]//Empirical Methods in Natural Language Processing, 2015: 1412-1421. [8] NGUYEN T H, GRISHMAN R. Relation extraction: Perspective from convolutional neural networks[C]//Proceedings of the 1st Workshop on Vector Space Modeling for Natural Language Processing, 2015: 39-48. [9] ZHANG D, WANG D. Relation classification via recurrent neural network[J]. CoRR abs/1508.01006, 2015. [10] HASHIMOTO K, MIWA M, TSURUOKA Y, et al. Simple customization of recursive neural networks for semantic relation classification[C]//Proceedings of the 2013 Conference on Empirical Methods in Natural Language Processing, 2013: 1372-1376. [11] EBRAHIMI J, DOU D. Chain based RNN for relation classification[C]//Proceedings of the 2015 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, 2015: 1244-1249. [12] SUNDERMEYER M, SCHLÜTER R, NEY H. LSTM neural networks for language modeling[C]//Thirteenth annual conference of the international speech communication association, 2012. [13] XU Y, MOU L, LI G, et al. Classifying relations via long short term memory networks along shortest dependency paths[C]//Proceedings of the 2015 Conference on Empirical Methods in Natural Language Processing, 2015: 1785-1794. [14] HOCHREITER S, SCHMIDHUBER J. Long short-term memory[J]. Neural computation, 1997, 9(8):1735-1780. doi: 10.1162/neco.1997.9.8.1735 [15] ZENG D, LIU K, LAI S, et al. Relation classification via convolutional deep neural network[C]//25th International Conference on Computational Linguistics, Proceedings of the Conference: Technical Papers, August 23-29, 2014: 2335-2344. [16] DAUPHIN Y N, FAN A, AULI M, et al. Language modeling with gated convolutional networks[C]//Proceedings of the 34th International Conference on Machine Learning-Volume 70, 2017: 933-941. [17] KINCHIN I M, HAY D B, ADAMS A. How a qualitative approach to concept map analysis can be used to aid learning by illustrating patterns of conceptual development[J]. Educational research, 2000, 42(1):43-57. doi: 10.1080/001318800363908 [18] BAHDANAU D, CHO K, BENGIO Y. Neural machine translation by jointly learning to align and translate[C]//International Conference on Learning Representations, 2015. [19] ZENG D, LIU K, CHEN Y, et al. Distant supervision for relation extraction via piecewise convolutional neural networks[C]//Proceedings of the 2015 Conference on Empirical Methods in Natural Language Processing, 2015: 1753-1762. [20] MNIH V, HEESS N, GRAVES A. Recurrent models of visual attention[C]//Advances in neural information processing systems, 2014: 2204-2212.