Advertisement

The Research About Topic Extraction Method Based on the DTS-ILDA Model

  • Xiaoli Guo
  • Li Feng
  • Yuhan Sun
  • Ping Guo
Conference paper
Part of the Smart Innovation, Systems and Technologies book series (SIST, volume 109)

Abstract

Because the existing LDA model is difficult to determine the number of topics and the key point of time, it is difficult to explain the topic result accurately. In this paper, the DTS-ILDA model is proposed, which fused an improved clustering algorithm into the DTM model, and label information is used for supervised learning on each subset. The size of the sliding window varies according to the topic distribution characteristics in this model. Text segmentation can be achieved more reasonable. The number of topics is also variable and easy to understand. The experiment shows that this method can effectively find the time points of important changes in the topic content, and prevent insignificance topics. It can reduce the related interference of the wrong topics and dig out the exact deep relationship at the same time.

Keywords

Model Time window segmentation Label Dynamic supervision 

Notes

Acknowledgment

This work is supported by the Science and Technology Development Project of Jilin Province of China (20180201092GX); Science and Technology Development Project of Jilin Province of China (20180101335JC). The project of teaching reform of higher education in Jilin Province of China (NO. 4 in 2017).

References

  1. 1.
    Li, F.H., Zheng, D.Q., Zhao, T.J.: Dunamic incremental analysis of sub-topic evolution. J. Comput. Res. Dev. 5(11), 2441–2450 (2015)Google Scholar
  2. 2.
    Yan, Y.Y., Tao, Y.B., Lin, H.: Interactice topic modeling based on hierarchical dirichlet process. J. Softw. 34(5), 1114–1126 (2016)Google Scholar
  3. 3.
    Cui, W., Liu, S., Tan, L., et al.: Text flow: towards better understanding of evolving topics in text. IEEE Trans. Visual. Comput. Graph. 17(12), 2412–2421 (2011)Google Scholar
  4. 4.
    Meng, F.Q., Su, X.H., Qu, Z.Y.: Nonlinear approach for estimating WCET during programming phase, cluster computing. J. Netw. Softw. Tools Appl. 19(3), 1449–1459 (2016)Google Scholar
  5. 5.
    Meng, F.Q., Su, X.H., Qu, Z.Y.: Interactive WCET prediction with warning for timeout risk. Int. J. Pattern Recogn. Artif. Intell. 31(05), 1750012 (2017)CrossRefGoogle Scholar
  6. 6.
    Meng, F.Q., Su, X.H.: WCET optimization strategy based on source code refactoring. Cluster Comput. J. Netw. Softw. Tools Appl.Google Scholar
  7. 7.
    Qu, Z.Y., Fan, X.D., Qu, N., Yu, H.T.: Smart grid text knowledge acquisition model based onontology. J. Northeast Dianli. Univ. 34(5), 60–68 (2014)Google Scholar
  8. 8.
    Guo, X.L., Han, X.: Grid knowledge collaborative discovery strategy research. J Northeast Dianli Univ 34(1), 94–98 (2014)Google Scholar

Copyright information

© Springer Nature Switzerland AG 2019

Authors and Affiliations

  1. 1.School of Information Engineering of Northeast Electric Power UniversityJilin CityChina
  2. 2.State Grid Xin Yuan Fengman Training CenterJilin CityChina
  3. 3.Liaoning Jianzhu Vocational CollogeLiaoyang CityChina

Personalised recommendations