Active dialogue method and device and voice terminal
A voice terminal and active technology, applied in the computer field, can solve the problems of small correlation between dialogue and users, difficulty in establishing a robot trust relationship, obtrusiveness, etc., and achieve the effect of avoiding stickiness and low usage.
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
Method used
Image
Examples
Embodiment 1
[0066] see figure 1 , is a schematic flowchart of an active dialogue method provided in the embodiment of the present application. Such as figure 1 As shown, the active dialogue method mainly includes the following processes:
[0067] S101. Search for a target topic corresponding to a user's topic of interest according to a preset topic switching table.
[0068] The active dialogue method provided in this embodiment is applied to a voice terminal, such as a smart speaker, a voice robot, etc., to implement an active dialogue between a voice terminal and a user. It should be noted that the active dialogue here can be the voice terminal actively initiates a dialogue with the user, or it can be that the voice terminal makes an active reply that is highly relevant to the user according to the voice initiated by the user, not just the conventional Question-and-answer passive dialogue. The applicable scenario of this solution can be a family or a fixed place, that is, a scenario ...
specific Embodiment approach
[0076] According to a specific implementation manner of the present disclosure, the preset topic statistics method includes:
[0077] Segment the dialogue into words, and record the basic topic to which the segmented words belong;
[0078] Count the occurrence frequency of each basic topic;
[0079] The basic topic with the highest frequency will be used as the topic of the conversation.
[0080] In the dialogue, the topic is usually reflected by the topic type to which the word belongs. By analyzing the topic type of the word in the dialogue, it is counted whether there are different topics. After acquiring the above conversations including multiple chat sessions, the chat sentences in the conversations are word-segmented according to the order of the conversations, and the words are divided one by one, and then the subsequent topic statistics operation is performed.
[0081] When performing word segmentation, it can be segmented according to the number of bytes covered by ...
Embodiment approach
[0123] In addition, according to another specific implementation manner of the present disclosure, the step of finding the target sentence with the highest semantic matching score from all the standby sentences includes:
[0124] According to the preset topic statistics method, the actual topic of each spare sentence is extracted;
[0125] If there is a first-class standby sentence whose actual topic belongs to any target topic;
[0126] Obtain the score value of the semantic matching between each of the first type of standby sentences and each target topic;
[0127] The backup sentence with the largest screening score is used as the target sentence.
[0128] In this embodiment, the generated 15 spare sentences are first marked with actual topics, compared with the top three target topics retained in the previous steps, and the spare sentences whose actual topics are not included in these three types of target topics are eliminated.
[0129] If there are still some spare sen...
PUM
Login to View More Abstract
Description
Claims
Application Information
Login to View More