Song retrieval method and system

By establishing a song search system, using multi-level search of keywords and attributes combined with user characteristics, the problem of low matching between search results and user needs in the existing technology is solved, and fast and accurate song matching and recommendation are achieved, improving user experience.

CN120067387APending Publication Date: 2025-05-30BEIJING THUNDERSTONE TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202510132365.8
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2025-02-06
Publication Date
2025-05-30

AI Technical Summary

Technical Problem

The existing music retrieval technology cannot accurately match user needs, resulting in users requiring multiple attempts to find the target song, and the inability to make specific recommendations based on specific users, resulting in poor user experience.

Method used

Multi-level search is performed through the keywords and attributes entered by the user, and fast and accurate song matching is performed in combination with user characteristics. The specific method includes establishing a mapping relationship table between the first word information and the entity ID and the mapping relationship table between the attribute and the song ID, searching and matching according to the search conditions entered by the user, and calculating the priority of the song for recommendation.

Benefits of technology

It improves the accuracy and efficiency of song retrieval, meets the diverse needs of different users, improves user experience, and makes song recommendations more in line with users' personal preferences.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN120067387A_ABST
    Figure CN120067387A_ABST
Patent Text Reader

Abstract

The embodiment of the invention discloses a song retrieval method and system, and the method comprises the steps: obtaining the song information of a song; establishing a first mapping relation table of the first word information and entity IDs according to the first word information in the song information, and establishing a second mapping relation table of attributes and song IDs according to attributes in the song information; when a retrieval condition input by a user is received, performing retrieval and matching in the first mapping relation table according to keywords in the retrieval condition to obtain a first song list; and / or retrieving in a second mapping relation table according to song attributes in the retrieval conditions to obtain a second song list; solving the intersection of the first song list and the second song list to obtain a target song list, and calculating the priority of each song in the target song list according to the song information and the user portrait of the user; and arranging and displaying the songs in the target song list according to the priority from high to low. And the retrieval efficiency and the user experience are improved.
Need to check novelty before this filing date? Find Prior Art

Claims

1. A song retrieval method, characterized in that: include: Get song information of a song; A first mapping relationship table between the first character information and the entity ID is established according to the first character information in the song information, and a second mapping relationship table between the attribute and the song ID is established according to the attribute in the song information; wherein the first character information includes the first character and the first letter, and the entity ID includes the song ID and the singer ID; When receiving the search condition input by the user, searching and matching in the first mapping relationship table according to the keywords in the search condition to obtain a first song list; and / or searching in the second mapping relationship table according to the song attributes in the search condition to obtain a second song list; When only the first song list or the second song list exists, the first song list or the second song list is used as the target song list; otherwise, the intersection of the first song list and the second song list is obtained to obtain the target song list, and the priority of each song in the target song list is calculated according to the song information and the user portrait of the user; The songs in the target song list are arranged and displayed in order of priority from high to low.

2. The song retrieval method according to claim 1, characterized in that: The first mapping relationship table between the first character information and the entity ID is established according to the first character information in the song information, including: Extract the first letter and the first letter of the song title from the song information; Extract the first letter and initial of the singer's name from the song information; The first letter of the song name, the first letter of the song name, the first letter of the singer name, and the first letter of the singer name are used as keys, and the corresponding entity ID is used as the value to generate a key-value pair; All key-value pairs are stored in a hash table to obtain the first mapping relationship table.

3. The song retrieval method according to claim 2, characterized in that: The second mapping relationship table between attributes and song IDs is established according to the attributes in the song information, including: Extracting song-related attributes from song information; Generate a key-value pair using the attribute as a key and the corresponding song ID as a value; All key-value pairs are stored in a hash table to obtain a second mapping relationship table.

4. The song retrieval method according to claim 3, characterized in that: The searching and matching in the first mapping relationship table according to the keywords in the search condition includes: Search the corresponding mapping relationship table according to the first word or first letter of the keyword to obtain an initial song list; Calculating the edit distance between the keyword and the entity name in the initial song list, deleting songs in the initial song list whose edit distance is greater than a preset distance threshold, and obtaining a filtered song list; Perform string fuzzy matching on the keywords and the entity names in the filtered song list to obtain the first song list.

5. The song retrieval method according to any one of claims 1 to 4, characterized in that: The priority of each song in the song list is calculated according to the song information and the user portrait of the user, and is calculated by the following formula: priority(song) =ω h *hot(song)+ω v *version(song)+ω n *isNew(song)+ω p *userProfile(song,userProfile)+oh c *isVIP(song)+ω s *score(song)+ω y *isYYB(song) Among them, priority(song) is the priority of the song, hot(song) is the popularity score, version(song) is the version score, isNew(song) represents whether the song is a new song, userProfile(song,userProfile) represents the user matching score, isVIP(song) represents whether the song is a VIP song, score(song) represents the quality score of the song, isYYB(song) represents whether the song is a transliterated version, ω h ,ω v ,ω n ,ω p ,ω c ,ω s ,ω y are all weight coefficients.

6. A song retrieval system, characterized in that: include: A data acquisition module, used to acquire song information of a song; A mapping relationship establishment module, used to establish a first mapping relationship table between the first character information and the entity ID according to the first character information in the song information, and to establish a second mapping relationship table between the attribute and the song ID according to the attribute in the song information; wherein the first character information includes the first character and the first letter, and the entity ID includes the song ID and the singer ID; A song matching module, for, when receiving a search condition input by a user, searching and matching in the first mapping relationship table according to the keywords in the search condition to obtain a first song list; and / or searching in the second mapping relationship table according to the song attributes in the search condition to obtain a second song list; A priority calculation module, for, when only the first song list or the second song list exists, using the first song list or the second song list as the target song list; otherwise, finding the intersection of the first song list and the second song list to obtain the target song list, and calculating the priority of each song in the target song list according to the song information and the user portrait of the user; The display module is used to arrange and display the songs in the target song list in order of priority from high to low.

7. The song retrieval system according to claim 6, characterized in that: The mapping relationship establishment module includes: The first first word extraction submodule is used to extract the first word and the first letter of the song title from the song information; The second first character extraction submodule is used to extract the first character and the first letter of the singer's name from the song information; A first key-value pair generation submodule is used to generate a key-value pair by using the first letter of a song title, the first letter of a song title, the first letter of a singer title, and the first letter of a singer title as keys and corresponding entity IDs as values; The first storage submodule is used to store all key-value pairs through a hash table to obtain the first mapping relationship table.

8. The song retrieval system according to claim 7, characterized in that: The mapping relationship establishment module includes: An attribute extraction submodule is used to extract song-related attributes from song information; A second key-value pair generation submodule is used to generate a key-value pair using the attribute as a key and the corresponding song ID as a value; The second storage submodule is used to store all key-value pairs through a hash table to obtain a second mapping relationship table.

9. The song retrieval system according to claim 8, characterized in that: The song matching module includes: A search submodule, for searching the corresponding mapping relationship table according to the first word or first letter of the keyword to obtain an initial song list; A filtering submodule, configured to calculate the edit distance between the keyword and the entity name in the initial song list, and delete songs in the initial song list whose edit distance is greater than a preset distance threshold, to obtain a filtered song list; The matching submodule is used to perform string fuzzy matching between the keyword and the entity name in the filtered song list to obtain the first song list.

10. The song retrieval system according to claim 9, characterized in that: The priority of each song in the target song list is calculated according to the song information and the user portrait by the following formula: priority(song) =ω h *hot(song)+ω v *version(song)+ω n *isNew(song)+ω p *userProfile(song,userProfile)+oh c *isVIP(song)+ω s *score(song)+ω y *isYYB(song) Among them, priority(song) is the priority of the song, hot(song) is the popularity score, version(song) is the version score, isNew(song) represents whether the song is a new song, userProfile(song,userProfile) represents the user matching score, isVIP(song) represents whether the song is a VIP song, score(song) represents the quality score of the song, isYYB(song) represents whether the song is a transliterated version, ω h ,ω v ,ω n ,ω p ,ω c ,ω s ,ω y are all weight coefficients.