基于深度学习的算法知识实体识别与发现

Identification discovery of algorithmic knowledge entity based on deep learning

随着互联网技术的快速发展,人类已经习惯于从网络上获取知识,然而伴随着网络资源爆炸式增长,网络资源内容多样,人们使用浏览器获取知识的方法却停滞不前,因此需要一种工具来帮助人们从网络中高效地获取和发现新知识。由于网络资源文本并不是完全结构化的数据,还包括一些自由文本等复杂的无结构数据,这种文本信息虽然方便人们自由表达概念以及事件等,但是同时也为机器搜索、统计分析等制造了障碍。因此,为了在文本上更方便地进行知识分析和挖掘,本文提出一种基于深度学习的算法知识实体识别与发现的方法,应用于算法知识领域来解决上述问题。通过创建算法知识专家库[1],训练词向量,建立深度神经网络模型,从算法知识文本中识别和发现算法知识名称。实验结果表明,该深度神经网络模型识别算法知识的准确率高达98%,并有效发现了专家库以外的新知识点,实现了预期实验需求。

With the rapid development of Internet technology, mankind has used to get from the network knowledge, however, along with the explosive growth of network resources, network resources showing diverse content, people use browser acquiring knowledge has become stagnant, and therefore a need is emerging for a tool to help people acquire and realize efficient discovery of new knowledge from the network. Since the fully structured data netw-ork resources are not text, also including complex unstructured data for some free text and so on. Although this text inlonnation is easy for people to freely express concepts and ev ents, while obstacles are formed for the machine search es, statistical analysis and others. Therefore, in order to more easily analyze and mining know-ledge on the text, the paper proposes algorithm knowledge entity recognition method based on the depth of learning, which is applied to the field of knowledge to solve the problem. By creating a pool of experts knowledge algorithm, training the word vector, build depth neural network model, achieve algorithm knowledge name recognition and discovery from the algorithm knowledge text. Experimental results show that the accuracy of the neural network model identification algorithm depth knowledge has been up to 98%, and effectively finds a new expert database knowledge so as to achieve the desired experimental needs.