METHODS AND SYSTEMS FOR DETERMINING RANKING OF DIALOGUE STICKER BASED ON SITUATION AND PREFERENCE INFORMATION

• US 20160210963A1
• Filed: 07/14/2015
• Published: 07/21/2016
• Est. Priority Date: 01/19/2015
• Status: Active Grant

• Current Assignee: NCSoft Corporation
• Original Assignee: NCSoft Corporation
• Inventors: Kim, Taek Jin, Lee, Jay June, Jang, Jungsun, Chung, Sehee, Lee, Kyeong Jong, Lee, Yeonsoo
• Granted Patent: US 9,792,903 B2
• Time in Patent Office: Days
• US Class Current: 1/1
• CPC Class Codes: G10L 15/1822 Parsing for meaning understanding
• CPC Class Codes: G10L 2015/088 Word spotting

Abstract

Provided is a method of determining the ranking of dialogue stickers based on situation and preference information. The method includes: analyzing utterances between the first user terminal and the second user terminal and selecting a sticker candidate group to be recommended to the first user terminal; calculating an expression suitability based on the preference and situation information in regard to the selected sticker candidate group; and determining the ranking of stickers in the order of suitability and then recommending a higher ranked sticker to the first user terminal.

Claims

1. A method of recommending a dialogue sticker based on situation and preference information, comprising:

by a server connected to a first user terminal and a second user terminal through a network,

analyzing utterances between the first user terminal and the second user terminal and selecting a sticker candidate group to be recommended to the first user terminal;

calculating an expression suitability based on the preference and situation information in regard to the selected sticker candidate group; and

determining the ranking of stickers in the order of suitability and then recommending a higher ranked sticker to the first user terminal.

2. The method of claim 1, wherein the calculating of the expression suitability comprises calculating a distance between an expression preference vector of a first user and an expression vector for each sticker in regard to each sticker of the selected sticker candidate group upon calculation of a preference-based expression suitability.

3. The method of claim 2, before the calculating of the expression suitability, further comprising generating the expression preference vector of the first user by accumulating and analyzing the utterances of the first user terminal.

4. The method of claim 1, wherein the calculating of the expression suitability comprises determining a similar situation to the utterances of the first user terminal and the second user terminal from a dialogue situation information database and calculating a distance between a sticker expression preference vector of the determined similar situation and expression vectors of each sticker of the selected sticker candidate group upon calculation of a situation-based expression suitability.

5. The method of claim 4, comprising determining the similar situation by retrieving situations having continuous utterances matching with a dialogue act flow of the utterances of the first user terminal and the second user terminal from the dialogue situation information database and determining the ranking of candidate situations in accordance with predefined dialogue act flow prediction criteria in regard to a retrieval result.

6. The method of claim 2, wherein the expression preference vector is one-dimensional vector having n vector elements, and each vector element is a criterion for determining the preference for a sticker.

7. The method of claim 6, wherein the vector element of the expression preference vector comprises at least one element of dynamic expression, complicated expression and exaggerated expression.

8. The method of claim 6, wherein the expression vector for each sticker is one-dimensional vector having the same vector elements as the expression preference vector, and is assigned and stored with a predetermined value for each sticker.

9. The method of claim 1, wherein the calculating of the expression suitability comprises inferring a relationship of a first user and a second user through analysis on utterances between the first user terminal and the second user terminal and correcting the expression suitability in accordance with the inferred relationship.

10. The method of claim 9, wherein when the intimacy between the first user and the second user is high, the dynamic expression, complicated expression or exaggerated expression suitability of the expression vector for each sticker is corrected so as to increase.

11. The method of claim 9, wherein when the relationship between the first user and the second user is a relationship between subordinates and superiors, the dynamic expression, complicated expression or exaggerated expression suitability of the expression vector for each sticker is corrected so as to decrease.

12. The method of claim 1, wherein the calculating of the expression suitability comprises calculating a recommendation score in which a preference-based expression suitability and a situation-based expression suitability for the selected sticker candidate group are parameters, and

the recommending of the higher ranked sticker to the first user terminal comprises recommending a sticker in an order of the recommendation score.

13. A computer program recorded in a recording medium comprising commands for executing:

analyzing utterances between the first user terminal and the second user terminal and selecting a sticker candidate group to be recommended to the first user terminal;

calculating an expression suitability—

in this case, calculating a distance between an expression preference vector of a first user and an expression vector for each sticker in regard to each sticker of the selected sticker candidate group upon calculation of a preference-based expression suitability, determining a similar situation to the utterances of the first user terminal and the second user terminal from a dialogue situation information database, and calculating a distance between a sticker expression preference vector of the determined similar situation and expression vectors of each sticker of the selected sticker candidate group upon calculation of a situation-based expression suitability—

based on preference or situation in regard to the selected sticker candidate group; and

determining the ranking of stickers in the order of suitability and then recommending a higher ranked sticker to the first user terminal.

14. The computer program of claim 13, wherein the calculating of the expression suitability comprises inferring a relationship of a first user and a second user through analysis on utterances between the first user terminal and the second user terminal and correcting the expression suitability in accordance with the inferred relationship.