Gesture query intention prediction method based on hidden Markov model

Abstract

The invention relates to a gesture query intention prediction method based on a hidden Markov model. Beginning from a user gesture interaction process, common gesture categories and corresponding feature attributes in gesture search are analyzed, then a gesture query intention model is established based on the hidden Markov characteristics, parameters of the gesture query intention model are calculated by using gesture interaction features on the basis, and optimal query intention corresponding to a gesture interaction event by using the viterbi theory. The query intention of the user is captured by using the gesture operation characteristics according to the fact that the gesture interaction process satisfies the characteristics of the hidden Markov process, and thus certain guidance is provided for the gesture interaction process of the user. The gesture query intention prediction method can be applied to search scenes facing to gesture interaction, therefore the user obtains information satisfying the attention in the gesture interaction process of the user, and the smoothness and degree of satisfaction of the gesture interaction are improved.

Description

technical field [0001] The invention relates to a gesture query intention prediction method based on a hidden Markov model, and belongs to the technical fields of human-computer interaction and information retrieval. Background technique [0002] With the rapid development of touch interaction technology, according to statistics, by 2017, 87% of the world's intelligent networked devices are tablets and smartphones, and the share of PCs is less than 13%, which shows that multi-touch interactive devices will become the most important part of human-computer interaction. Gesture interaction has gradually become one of the leading ways of human-computer interaction. Therefore, the user uses gestures such as sliding up, sliding down, zooming in, zooming out, and clicking to search for information during the information search process. In order to deepen the understanding of the searched content, the user repeatedly performs the above gestures and iteratively tries and errors. Howe...

AI Technical Summary

Problems solved by technology

This prediction method analyzes the user's click behavior on the search results corresponding to the original search statement based on the log information of the user's interaction with the search engine. It mainly considers the common click behavior and does not consider the application requirements of gesture interaction.

In another invention patent "Touch to Search" (Application No.: 201380072159.8, Authorization No.: CN 104969164 A), the user can select the content displayed on the touch screen by using gestures instead of typing a search query into the search interface, based on gesture data Identify a subset of content to select a set of candidate search queries, and calculate the likelihood score for each candidate search query, and finally select one or more candidate search queries. Although this invention makes full use of gesture data, it does not use gesture data Combined with user query intent

[0005] In summary, most methods do not focus on the relationship between gesture operation features and user query intentions, and extract information that meets user intentions.

Images