Electronic device, contents searching system and searching method thereof

1. A method for searching contents, the method comprising:

creating a content index using a natural language processing-based searching model to index content;

creating a keyword vector by extracting a keyword from a search input;

obtaining a candidate contents group based on a first similarity between the content index and the keyword vector;

creating a content vector for the candidate contents group using an artificial intelligence model trained using learning data from a user search log in which queries and contents corresponding to the respective queries are paired;

obtaining a second similarity between a query vector created by the artificial intelligence model based on the search input and the content vector; and

aligning and providing contents included in the candidate contents group based on the first similarity and the second similarity.

2. The method of claim 1 , further comprising:

obtaining text comprising information of at least one of a title, genre, actor, director, synopsis, or opening date of contents;

extracting a keyword from the obtained text;

obtaining term frequency and inverse document frequency of the extracted keyword; and

creating the natural language processing-based searching model by obtaining the content index based on the obtained term frequency and the inverse document frequency.

3. The method of claim 2 , wherein the content index comprises a plurality of fields, and

wherein the obtaining the candidate contents group comprises assigning a weight to at least one field among the plurality of fields.

4. The method of claim 1 , further comprising:

obtaining text comprising information of at least one of a title, genre, actor, director, synopsis, or opening date of contents;

creating a plurality of sentences based on a keyword extracted from the obtained text;

calculating a score based on a similarity between the plurality of created sentences;

identifying a preset number of sentences in which the calculated score is at least as high as a preset threshold;

creating the preset number of identified sentences as vector data of the contents; and

training the artificial intelligence model based on learning data comprising the vector data of the contents.

5. The method of claim 4 , wherein the training of the artificial intelligence model comprises:

adding a position embedding value to maintain an order of a keyword included in the vector data of the contents; and

inputting, to the artificial intelligence model, learning data connecting the vector data of the contents and the queries.

6. The method of claim 1 , wherein the creating of the keyword vector comprises, based on the search input being a voice:

identifying an error of recognized text based on similarity between text recognized in the voice and prestored query data; and

modifying an error of the recognized text.

7. An electronic device comprising:

an input interface comprising circuitry configured to receive a search input;

an output interface comprising output circuitry; and

memory storing instructions, configured to be executed by one or more processors, for controlling the electronic device to perform operations comprising:

creating a content index using a natural language processing-based searching model to index content;

creating a keyword vector by extracting a keyword from the search input,

obtaining a candidate contents group based on a first similarity between the content index and the keyword vector,

creating a content vector for the candidate contents group using an artificial intelligence model trained using learning data from a user search log in which queries and contents corresponding to the respective queries are paired,

obtaining a second similarity between a query vector created by the artificial intelligence model based on the search input and the content vector,

aligning contents included in the candidate contents group based on the first similarity and the second similarity, and

controlling the output interface to provide the aligned contents.

8. The electronic device of claim 7 , comprising memory storing instructions, configured to be executed by one or more processors, for controlling the electronic device to perform operations comprising:

obtaining text comprising information of at least one of a title, genre, actor, director, synopsis, or opening date of contents, extracting a keyword from the obtained text, obtaining term frequency and inverse document frequency of the extracted keyword, and creating the natural language processing-based searching model by obtaining the content index based on the obtained term frequency and the inverse document frequency.

9. The electronic device of claim 8 , wherein the content index comprises a plurality of fields, and

wherein the electronic device comprises memory storing instructions, configured to be executed by one or more processors, for controlling the electronic device to perform operations comprising assigning a weight to at least one field among the plurality of fields.

10. The electronic device of claim 7 , comprising memory storing instructions, configured to be executed by one or more processors, for controlling the electronic device to perform operations comprising:

obtaining text comprising information of at least one of a title, genre, actor, director, synopsis, or opening date of contents,

creating a plurality of sentences based on a keyword extracted from the obtained text,

calculating a score based on a similarity between the plurality of created sentences,

identifying a preset number of sentences in which the calculated score is high,

creating the preset number of identified sentences as vector data of the contents, and

training the artificial intelligence model based on learning data comprising the vector data of the contents.

11. The electronic device of claim 10 , comprising memory storing instructions, configured to be executed by one or more processors, for controlling the electronic device to perform operations comprising:

adding a position embedding value to maintain an order of a keyword included in the vector data of the contents; and

inputting, to the artificial intelligence model, learning data connecting the vector data of the contents and the queries.

12. The electronic device of claim 7 , comprising memory storing instructions, configured to be executed by one or more processors, for controlling the electronic device to perform operations comprising, based on the search input being a voice:

identifying an error of recognized text based on similarity between text recognized in the voice and prestored query data; and

modifying an error of the recognized text.

13. A contents searching system comprising a terminal device and a server, wherein

the terminal device comprises circuitry configured to receive a search input and transmit the search input to the server; and

the server is configured to receive the search input from the terminal device,

wherein the server is further configured to:

create a content index using a natural language processing-based searching model to index content;

create a keyword vector by extracting a keyword from the search input,

obtain a candidate contents group based on a first similarity between the content index and the keyword vector,

create a content vector for the candidate contents group using an artificial intelligence model trained using learning data from a user search log in which queries and contents corresponding to the respective queries are paired,

obtain a second similarity between a query vector created by the artificial intelligence model based on the search input and the content vector,

align contents included in the candidate contents group based on the first similarity and the second similarity, and

transmit the aligned contents to the terminal device,

wherein the terminal device is further configured to output the aligned contents received from the server.