Search

In this paper, two different technologies are adopted to realize the digital campus intelligent chatbot. The robot has three core modules: chat module, campus information module and virtual teaching assistant module. For chat module, this paper uses Encoder-Decoder framework. For the campus information module and virtual teaching assistant module, this paper pursue high precision and adopt retrieval mode. In practice, the two modules implemented by different technical means complement each other, greatly improving the user experience of students.

This paper focuses on the Bloom Filter deformation algorithm. The storage space occupied by Bloom Filter is independent of elements, which has high space efficiency and low time complexity of insert and query operations. However, as the number of added collections increases, the Bloom Filter’s error rate increases, thus filtering out a lot of non-repeating information. In order to solve this problem, the standard Bloom Filter is improved by introducing an algorithm of the structure of the master-slave Bloom Filter. Only when the master-slave Bloom Filter produces errors, the filter is considered as an error. At the same time, by dynamically increasing the number of filters, the growth of misjudgment rate is delayed. This paper also involves an improved Bloom Filter algorithm to reduce the error rate of the repeated data in the judgment filter.