Personalized recommendation method and device and storage medium

Abstract

The invention relates to a personalized recommendation method and device and a storage medium. The method comprises the steps of obtaining historical voice data of a user, and obtaining first pronunciation information of the user based on the historical voice data; determining regional information of the user according to the first pronunciation information and second pronunciation information corresponding to the plurality of regional accents; and recommending candidate objects to the user based on the regional information of the user. According to the embodiment of the invention, the regional information of the user can be deduced from the accent in the voice data of the user, and the regional information of the user can be determined more accurately because the voice contains rich dialects or accent information, so that the user experience is improved. And the conditions of the mixed accent and the mixed region can be considered through the second pronunciation information corresponding to the multiple region accent, and more comprehensive region information is obtained, so that the recommendation effect can be more comprehensive.

Description

Technical Field

[0001] The present disclosure relates to the field of artificial intelligence technology, and in particular to a personalized recommendation method, device and storage medium. Background Art

[0002] Personalized recommendation systems can provide users with the most relevant and personalized information, product or service recommendations by analyzing user behavior and preferences, and are widely used in many fields. User regional information plays an important role in personalized recommendations. Regional information can help the system understand the user's possible language and cultural preferences, thereby providing recommendations that are closer to the user.

[0003] However, in current personalized recommendation systems, users are usually asked to provide their geographic location information when registering, but they may be unwilling to provide such information or unable to provide accurate and up-to-date information. Existing technologies that use deep learning models to predict speakers' geographic information require a large amount of data for training, and have poor discrimination accuracy. They cannot solve the problem of mixed accents and mixed regions, resulting in incomplete recommendations. Summary of the invention

[0004] In view of this, the present disclosure proposes a personalized recommendation method, device and storage medium.

[0005] According to one aspect of the present disclosure, a personalized recommendation method is provided. The method comprises:

[0006] Acquire historical voice data of the user, and obtain first pronunciation information of the user based on the historical voice data;

[0007] Determining the user's regional information based on the first pronunciation information and the second pronunciation information corresponding to the plurality of regional accents;

[0008] Recommend candidate objects to users based on their geographical information.

[0009] In a possible implementation, obtaining the first pronunciation information of the user based on the historical voice data includes:

[0010] Perform text transcription and pinyin transcription on the historical voice data to obtain a text sequence of the historical voice data and a pinyin syllable sequence pronounced by the user;

[0011] Determine a Mandarin pinyin sequence based on a text sequence;

[0012] Based on the association between the text sequence, the Mandarin Pinyin sequence and the Pinyin syllable sequence pronounced by the user, a triple corresponding to each Chinese character text in the text sequence is obtained.

[0013] In a possible implementation, obtaining the first pronunciation information of the user based on the historical voice data further includes:

[0014] Among the triples corresponding to each Chinese character text in the text sequence, the triples whose Mandarin pronunciation is the same as the user's pronunciation are removed.

[0015] In a possible implementation, the second pronunciation information includes a first association relationship between a Chinese character text and a pronunciation in a corresponding regional accent, and a second association relationship between a Mandarin pronunciation and a pronunciation in a corresponding regional accent. Determining the user's regional information according to the first pronunciation information and the second pronunciation information corresponding to the plurality of regional accents includes:

[0016] The user's region information is determined based on the first pronunciation information and the first association relationship, and the first pronunciation information and the second association relationship.

[0017] In a possible implementation, the first pronunciation information includes a plurality of triples, each triple includes a Chinese character text, a Mandarin pronunciation, and a user pronunciation, and the user's regional information is determined according to the first pronunciation information and the first association relationship, and the first pronunciation information and the second association relationship, including:

[0018] For each triple in the plurality of triples, matching the Chinese text and the user pronunciation in the triple with the first association relationships corresponding to the plurality of regions respectively to obtain a first matching result;

[0019] For each triple in the plurality of triples, matching the Mandarin pronunciation and the user pronunciation in the triple with the second association relationships corresponding to the plurality of regions respectively to obtain a second matching result;

[0020] Based on the first matching result and the second matching result, the region information of the user is determined.

[0021] In a possible implementation, determining the user's region information based on the first matching result and the second matching result includes:

[0022] In response to the first matching result being a successful match, adding points to the corresponding regional tag;

[0023] In response to the second matching result being a successful match, adding points to the corresponding regional tag;

[0024] The regional tags and their scores are used as the user's regional information.

[0025] In a possible implementation, the method further includes:

[0026] Determine multiple regional tags based on regional information of pre-collected data;

[0027] Regional accent pronunciation libraries are collected for each of the multiple regional tags, and each regional accent pronunciation library stores second pronunciation information of the corresponding region.

[0028] In a possible implementation, regional accent pronunciation libraries are collected for multiple regional tags respectively, including:

[0029] Collect pronunciation information of regional accents corresponding to each regional label that is different from Mandarin;

[0030] Based on the pronunciation information of the regional accent that is different from that of Mandarin, a key-value pair representing a first association relationship and a key-value pair representing a second association relationship are determined; the key of the key-value pair representing the first association relationship is a Chinese text, and the value is the pronunciation of the corresponding regional accent; the key of the key-value pair representing the second association relationship is a Mandarin pronunciation, and the value is the pronunciation of the corresponding regional accent;

[0031] Based on the key-value pair indicating the first association relationship and the key-value pair indicating the second association relationship, a regional accent pronunciation library is determined.

[0032] According to another aspect of the present disclosure, a personalized recommendation device is provided. The device includes:

[0033] A first determination module, used to obtain historical voice data of the user, and obtain first pronunciation information of the user based on the historical voice data;

[0034] A second determination module, configured to determine the user's regional information based on the first pronunciation information and second pronunciation information corresponding to a plurality of regional accents;

[0035] The recommendation module is used to recommend candidate objects to users based on their geographical information.

[0036] In a possible implementation manner, the first determining module is configured to:

[0037] Perform text transcription and pinyin transcription on the historical voice data to obtain a text sequence of the historical voice data and a pinyin syllable sequence pronounced by the user;

[0038] Determine a Mandarin pinyin sequence based on a text sequence;

[0039] Based on the association between the text sequence, the Mandarin Pinyin sequence and the Pinyin syllable sequence pronounced by the user, a triple corresponding to each Chinese character text in the text sequence is obtained.

[0040] In a possible implementation manner, the first determining module is further configured to:

[0041] Among the triples corresponding to each Chinese character text in the text sequence, the triples whose Mandarin pronunciation is the same as the user's pronunciation are removed.

[0042] In a possible implementation, the second pronunciation information includes a first association relationship between a Chinese character text and a pronunciation in a corresponding regional accent, and a second association relationship between a Mandarin pronunciation and a pronunciation in a corresponding regional accent, and the second determination module is used to:

[0043] The user's region information is determined based on the first pronunciation information and the first association relationship, and the first pronunciation information and the second association relationship.

[0044] In a possible implementation, the first pronunciation information includes a plurality of triples, each triple includes a Chinese character text, a Mandarin pronunciation, and a user pronunciation, and the user's regional information is determined according to the first pronunciation information and the first association relationship, and the first pronunciation information and the second association relationship, including:

[0045] For each triple in the plurality of triples, matching the Chinese text and the user pronunciation in the triple with the first association relationships corresponding to the plurality of regions respectively to obtain a first matching result;

[0046] For each triple in the plurality of triples, matching the Mandarin pronunciation and the user pronunciation in the triple with the second association relationships corresponding to the plurality of regions respectively to obtain a second matching result;

[0047] Based on the first matching result and the second matching result, the region information of the user is determined.

[0048] In a possible implementation, determining the user's region information based on the first matching result and the second matching result includes:

[0049] In response to the first matching result being a successful match, adding points to the corresponding regional tag;

[0050] In response to the second matching result being a successful match, adding points to the corresponding regional tag;

[0051] The regional tags and their scores are used as the user's regional information.

[0052] In a possible implementation, the device further includes:

[0053] A third determination module, configured to determine a plurality of region tags according to the region information of the pre-collected data;

[0054] The collection module is used to collect regional accent pronunciation libraries for multiple regional tags respectively, and each regional accent pronunciation library stores the second pronunciation information of the corresponding region.

[0055] In a possible implementation, the collection module is used to:

[0056] Collect pronunciation information of regional accents corresponding to each regional label that is different from Mandarin;

[0057] Based on the pronunciation information of the regional accent that is different from that of Mandarin, a key-value pair representing a first association relationship and a key-value pair representing a second association relationship are determined; the key of the key-value pair representing the first association relationship is a Chinese text, and the value is the pronunciation of the corresponding regional accent; the key of the key-value pair representing the second association relationship is a Mandarin pronunciation, and the value is the pronunciation of the corresponding regional accent;

[0058] Based on the key-value pair indicating the first association relationship and the key-value pair indicating the second association relationship, a regional accent pronunciation library is determined.

[0059] According to another aspect of the present disclosure, a personalized recommendation device is provided, comprising: a processor; and a memory for storing processor-executable instructions; wherein the processor is configured to implement the above method when executing the instructions stored in the memory.

[0060] According to another aspect of the present disclosure, a non-volatile computer-readable storage medium is provided, on which computer program instructions are stored, wherein the computer program instructions implement the above method when executed by a processor.

[0061] According to another aspect of the present disclosure, a computer program product is provided, including a computer-readable code, or a non-volatile computer-readable storage medium carrying the computer-readable code. When the computer-readable code runs in a processor of an electronic device, the processor in the electronic device executes the above method.

[0062] According to an embodiment of the present application, the user's historical voice data is obtained to determine the user's regional information, so that the user does not need to actively provide the regional information. The user's first pronunciation information is obtained based on the historical voice data, and the user's regional information is determined based on the first pronunciation information and the second pronunciation information corresponding to multiple regional accents. Candidate objects are recommended to the user based on the regional information. It is possible to infer the user's regional information from the accent in the user's voice data. Since the voice contains rich dialect or accent information, the user's regional information can be determined more accurately. In addition, the present application can take into account the situation of mixed accents and mixed regions through the second pronunciation information corresponding to multiple regional accents, and obtain more comprehensive regional information, thereby making the recommendation effect more comprehensive.

[0063] Further features and aspects of the present disclosure will become apparent from the following detailed description of exemplary embodiments with reference to the attached drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] The accompanying drawings, which are incorporated in and constitute a part of the specification, illustrate exemplary embodiments, features, and aspects of the disclosure and, together with the description, serve to explain the principles of the disclosure.

[0065] Figure 1 A schematic diagram showing an application scenario according to an embodiment of the present application.

[0066] Figure 2 A flowchart of a personalized recommendation method according to an embodiment of the present application is shown.

[0067] Figure 3 A flowchart of a personalized recommendation method according to an embodiment of the present application is shown.

[0068] Figure 4 A structural diagram of a personalized recommendation device according to an embodiment of the present application is shown.

[0069] Figure 5 It is a block diagram of a device 1900 for personalized recommendation according to an exemplary embodiment. DETAILED DESCRIPTION

[0070] Various exemplary embodiments, features and aspects of the present disclosure will be described in detail below with reference to the accompanying drawings. The same reference numerals in the accompanying drawings represent elements with the same or similar functions. Although various aspects of the embodiments are shown in the accompanying drawings, the drawings are not necessarily drawn to scale unless otherwise specified.

[0071] The word “exemplary” is used exclusively herein to mean “serving as an example, example, or illustration.” Any embodiment described herein as “exemplary” is not necessarily to be construed as preferred or advantageous over other embodiments.

[0072] In addition, in order to better illustrate the present disclosure, numerous specific details are given in the following specific embodiments. It should be understood by those skilled in the art that the present disclosure can also be implemented without certain specific details. In some examples, methods, means, components and circuits well known to those skilled in the art are not described in detail in order to highlight the subject matter of the present disclosure.

[0073] Personalized recommendation systems can provide users with the most relevant and personalized information, product or service recommendations by analyzing user behavior and preferences, and are widely used in many fields. User geographical information plays an important role in personalized recommendations. Geographical information can help the system understand the user's possible language and cultural preferences, thereby providing recommendations that are closer to the user. However, in current personalized recommendation systems, users are usually asked to provide geographic location information when registering, but users may be unwilling to provide such information, or unable to provide accurate and up-to-date information. Existing technologies that use deep learning models to predict the speaker's geographical information require a large amount of data for training, and have poor discrimination accuracy. They cannot solve the problems of mixed accents and mixed regions, resulting in incomplete recommendation effects.

[0074] In view of this, the present application provides a personalized recommendation method, device and storage medium. The method of the embodiment of the present application can determine the user's regional information by obtaining the user's historical voice data, so that the user does not need to actively provide regional information, obtain the user's first pronunciation information based on the historical voice data, and determine the user's regional information based on the first pronunciation information and the second pronunciation information corresponding to multiple regional accents, and recommend candidate objects to the user based on the regional information, so that the user's regional information can be inferred from the accent in the user's voice data. Since the voice contains rich dialect or accent information, the user's regional information can be determined more accurately, and the present application can take into account the situation of mixed accents and mixed regions through the second pronunciation information corresponding to multiple regional accents, and obtain more comprehensive regional information, so that the recommendation effect can be more comprehensive.

[0075] Figure 1 The personalized recommendation system of the embodiment of the present application can be used in scenarios such as e-commerce websites, movie and music recommendation platforms, social media, etc., to recommend personalized information, products or service suggestions to users, or in recommendation scenarios such as intelligent recommendation modules with multiple rounds of voice interaction, such as Figure 1 As shown, the personalized recommendation system of the embodiment of the present application can collect the user's historical voice data, obtain the user's geographical information based on the user's historical voice data analysis (for example, analysis shows that the user is from Sichuan), and recommend candidate objects to the user based on the geographical information, the candidate objects include personalized information, product or service recommendations, etc.

[0076] The personalized recommendation system of the embodiment of the present application can also be used in other recommendation scenarios, and the present application does not limit this, as long as the user's voice data can be obtained in the scenario.

[0077] The personalized recommendation system of the embodiment of the present application can be deployed on a terminal device or a server. The terminal device can be any one or more of a mobile phone, a foldable electronic device, a tablet computer, a desktop computer, a laptop computer, a handheld computer, a notebook computer, an ultra-mobile personal computer (UMPC), a netbook, a cellular phone, a personal digital assistant (PDA), and a vehicle-mounted device. The embodiment of the present disclosure does not impose any special restrictions on the specific type of the terminal device, and it can have a wired or wireless communication function.

[0078] The server can be located locally or in the cloud, and can be a physical device or a virtual device, such as a virtual machine, a container, etc., and has a wireless communication function, wherein the wireless communication function can be set in the chip (system) or other parts or components of the server. The wireless communication function can be realized, for example, through mobile communication technologies such as 2G / 3G / 4G / 5G, as well as Wi-Fi, Bluetooth, frequency modulation (FM), digital radio, satellite communication, etc. Communication can also be carried out through wired connection to achieve interaction with other devices.

[0079] Figure 2 A flowchart of a personalized recommendation method according to an embodiment of the present application is shown. The method can be used in the above-mentioned personalized recommendation system, such as Figure 2 As shown, the method may include:

[0080] Step S201, acquiring historical voice data of a user, and obtaining first pronunciation information of the user based on the historical voice data.

[0081] The user's historical voice data may be the user's voice data saved before personalized recommendations are made for the user, and the user's historical voice data may be one or more sentences. The personalized recommendation system may include a user portrait module, and the user's historical voice data may be stored in the user portrait module.

[0082] Before step S201, it is also possible to:

[0083] Determine multiple regional tags according to the regional information of the pre-collected data; collect regional accent pronunciation libraries respectively for the multiple regional tags, and store the second pronunciation information of the corresponding region in each regional accent pronunciation library.

[0084] Among them, the pre-collected data may include personalized information, product or service recommendations, etc., which are not limited in this application. Multiple regional labels can be determined in advance based on the regional information of the pre-collected data, and each candidate object can be associated with one or more regional labels. For example, when the candidate object is a specialty of Province M, its regional label can be "Province M", and when the candidate object is an introduction to tourist attractions in City X Province Y, the regional label can be "Province X", or "City Y", or two regional labels of "Province X" and "City Y". The granularity of the regional label can include provinces, cities, counties, etc., and the granularity can be set and adjusted as needed. The same candidate object can correspond to a regional label of a preset granularity, or it can correspond to multiple regional labels of preset multiple granularities. The above-mentioned personalized recommendation system may include a candidate object module, and one or more regional labels of corresponding granularity can be set for each candidate object in the candidate object module.

[0085] Based on a determined plurality of regional tags (for example, there are N types in total, corresponding to N regions respectively), regional accent pronunciation libraries can be collected for the N types of regional tags respectively, and second pronunciation information corresponding to each region is stored in each regional accent pronunciation library. The second pronunciation information can be used to describe the pronunciation information different from Mandarin in the corresponding regional accent. For example, the pronunciation different from Mandarin in the accent of the Sichuan region can be stored in the regional accent pronunciation library corresponding to the regional tag "Sichuan".

[0086] The second pronunciation information may include a first association relationship between the Chinese character text and the pronunciation in the corresponding regional accent, and a second association relationship between the Mandarin pronunciation and the pronunciation in the corresponding regional accent.

[0087] Each regional accent pronunciation library can be hierarchically designed based on key-value pairs, and can be divided into two levels: Chinese characters and syllables. Each level can include multiple key-value pairs. Among them, the Chinese character level can represent the first association relationship between the Chinese character text and the pronunciation in the corresponding regional accent, and the syllable level can represent the second association relationship between the Mandarin pronunciation and the pronunciation in the corresponding regional accent.

[0088] In the process of collecting regional accent pronunciation libraries for a plurality of regional tags respectively, it is possible to: collect the pronunciation information different from Mandarin in the regional accents corresponding to each regional tag; determine the key-value pairs representing the first association relationship and the key-value pairs representing the second association relationship based on the pronunciation information different from Mandarin in the regional accent; and determine the regional accent pronunciation library based on the key-value pairs representing the first association relationship and the key-value pairs representing the second association relationship.

[0089] The key-value pairs representing the first association relationship are the key-value pairs in the Chinese character level, where the key is the Chinese character text and the value is the pronunciation in the corresponding regional accent. An example of one key-value pair is: (North-bei1), indicating that in the accent of the corresponding region, the Chinese character "North" will be pronounced as "bei1" fixedly.

[0090] The key-value pairs representing the second association relationship are the key-value pairs in the syllable level, where the key is the Mandarin pronunciation and the value is the pronunciation in the corresponding regional accent. An example of one key-value pair is: (bei3-bei1), indicating that in the accent of the corresponding region, the syllable "bei3" will be pronounced as "bei1" fixedly.

[0091] Thus, multiple regional accent pronunciation libraries can be constructed for multiple regions respectively.

[0092] The first pronunciation information obtained based on historical speech data may include multiple triples, each triple including Chinese character text, Mandarin pronunciation, and user pronunciation. The Chinese character text is, for example, "北 (north)"; the Mandarin pronunciation can be the standard Mandarin pinyin of the Chinese character text, for example, "bei3"; the user pronunciation can be the pinyin of the user's actual pronunciation of the Chinese character text in the corresponding historical speech data, for example, "bei1". An example of a triple thus formed is (北, bei3, bei1).

[0093] The following introduces the method of obtaining the user's first pronunciation information based on historical speech data. Refer to Figure 3 , which shows a flowchart of a personalized recommendation method according to an embodiment of the present application. As Figure 3 shown, in step S201, it may include:

[0094] Step S301: Transcribe the historical speech data into text and pinyin, obtaining the text sequence of the historical speech data and the pinyin syllable sequence of the user's pronunciation.

[0095] Among them, speech recognition technology can be used to transcribe the historical speech data into text, obtaining the Chinese character text of the historical speech data, such as "我爱去北京旅游 (I love to travel to Beijing)";

[0096] Speech recognition technology can also be used to transcribe the historical speech data into pinyin, obtaining the pinyin syllable sequence of the user's pronunciation corresponding to the historical speech data, such as "wo3 ai4 qu4 bei1 jing3 lv3 you4", indicating the user's actual pronunciation in the historical speech data. The above speech recognition technology can be Conformer, Hidden Markov Model (HMM), etc., and the present application does not limit this.

[0097] Step S302: Determine the Mandarin pinyin sequence based on the text sequence.

[0098] Among them, natural language processing technology can be used to transform the text sequence obtained after the above text transcription to obtain the corresponding Mandarin pinyin sequence. For example, the Mandarin pinyin sequence obtained after transforming the text sequence "我爱去北京旅游 (I love to travel to Beijing)" is "wo3 ai4 qu4 bei3 jing1 lv3 you2". Natural language processing technology such as word segmentation, G2P (grapheme-to-Phoneme), etc. are used, and the present application does not limit this.

[0099] Step S303: Obtain the triple corresponding to each Chinese character text in the text sequence based on the association relationship between the text sequence, the Mandarin pinyin sequence, and the pinyin syllable sequence of the user's pronunciation.

[0100] For example, for M sentences of historical speech data, M sentence-level triples can be obtained, where each triple includes (text sequence, Mandarin pinyin sequence, pinyin syllable sequence of the user's pronunciation). The M sentence-level triples can be expanded into character-level triples, where the text sequence in the sentence-level triple can be split into individual Chinese character texts, and the triples corresponding to each Chinese character text in the text sequence can be obtained. For example, for the text sequence "I love to travel in Beijing", 7 triples corresponding to Chinese character texts can be obtained, and an example of one triple is (North, bei3, bei1).

[0101] For M sentences of historical speech data, for example, X triples corresponding to Chinese character texts can be obtained.

[0102] In step S201, it is also possible to:

[0103] Remove the triples in which the Mandarin pronunciation and the user's pronunciation are the same in the triples corresponding to each Chinese character text in the text sequence.

[0104] For example, among the 7 triples corresponding to the text sequence "I love to travel in Beijing", the triples corresponding to the Chinese character texts "I", "love", "go", and "travel" have the same Mandarin pronunciation and the user's pronunciation, and these 4 triples can be filtered out and removed.

[0105] Return to see Figure 2 :

[0106] Step S202, determine the user's regional information according to the first pronunciation information and the second pronunciation information corresponding to multiple regional accents.

[0107] The user's regional information can be used to represent the region where the user comes from, such as Sichuan, Hunan, etc., and its granularity can be controlled by the granularity of the above regional labels.

[0108] In step S202, it is possible to:

[0109] Determine the user's regional information according to the first pronunciation information and the first association relationship, and the first pronunciation information and the second association relationship.

[0110] For example, a regional accent portrait can be initialized for the user. The regional accent portrait can consist of N key-value pairs. The key in each key-value pair can correspond to a regional label, and the value in each key-value pair represents the score of the regional label. The higher the score, the more the user's accent matches the regional label, indicating that the user is more likely to come from the region indicated by the regional label. When initializing, the value of each key-value pair is 0.

[0111] When determining the user's regional information, for example, a regional accent pronunciation library can be used to match the triples of the multiple Chinese character texts obtained above to determine the user's regional accent portrait.

[0112] Among them, it is possible to:

[0113] For each of the multiple triples, match the Chinese character text and the user's pronunciation in the triple with the first association relationships corresponding to multiple regions to obtain a first matching result;

[0114] For each of the multiple triples, match the Mandarin pronunciation and the user's pronunciation in the triple with the second association relationships corresponding to multiple regions to obtain a second matching result;

[0115] Based on the first matching result and the second matching result, determine the user's regional information.

[0116] If the first matching result is a successful match, it can indicate that the Chinese character text and the user's pronunciation match the corresponding key-value pairs at the Chinese character level in the regional accent pronunciation library; if the second matching result is a successful match, it can indicate that the Mandarin pronunciation and the user's pronunciation match the corresponding key-value pairs at the syllable level in the regional accent pronunciation library.

[0117] In the process of determining the user's regional information based on the first matching result and the second matching result, it is possible to: in response to the first matching result being a successful match, add points to the corresponding regional label; in response to the second matching result being a successful match, add points to the corresponding regional label; use the regional labels and the scores of the regional labels as the user's regional information.

[0118] For example, for the triple (北, bei3, bei1), if the Chinese character text and the user's pronunciation in it match the key-value pair (北 - bei1) at the Chinese character level in the regional accent pronunciation library corresponding to Sichuan, in the regional accent portrait, 1 point can be added to the value corresponding to the regional label "Sichuan";

[0119] For another example, for the triple (北, bei3, bei1), if the Mandarin pronunciation and the user's pronunciation in it simultaneously match the key-value pair (bei3 - bei1) at the syllable level in the regional accent pronunciation library corresponding to Hunan, in the regional accent portrait, 1 point can also be added to the value corresponding to the regional label "Hunan".

[0120] After scoring the key-value pairs in the regional accent portrait for the multiple triples using the above process, N scored key-value pairs: (regional label - score) can be obtained as the user's regional accent portrait, and this regional accent portrait can be used as the user's regional information.

[0121] By utilizing the first association relationship and the second association relationship in the regional accent pronunciation library to match the triplet, the user's accent features can be matched in a more fine-grained manner, the false recall problem in the existing classification algorithm can be solved, and the accuracy of the obtained regional accent information can be improved.

[0122] Step S203: recommending candidate objects to the user based on the user's region information.

[0123] The user and the candidate object may be associated with each other based on the regional accent portrait to perform personalized recommendation and recommend the candidate object to the user. The personalized recommendation system may include a recommendation algorithm module, and step S203 may be performed in the recommendation algorithm module.

[0124] This application does not limit the method of recommending candidate objects to users based on regional information. In one example, in step S203, you can:

[0125] Based on the user's geographical information, candidate objects with corresponding geographical labels are recommended to the user.

[0126] Recommendations based on regional tags can make the personalized recommendation process more efficient.

[0127] For example, based on the key-value pair information in the regional accent portrait, the value in the key-value pair can be used as the weight corresponding to each regional label, and for each candidate object, the regional label corresponding to the candidate object is weighted by the weight to obtain the score of each candidate object. In this way, the weight distribution of the regional accents of users with mixed regional accents can be quantified, making the recommended candidates more accurate and the user experience better.

[0128] One or more candidate objects with the highest scores may be selected as candidate objects recommended to the user on a high score first basis.

[0129] According to an embodiment of the present application, the user's historical voice data is obtained to determine the user's regional information, so that the user does not need to actively provide the regional information. The user's first pronunciation information is obtained based on the historical voice data, and the user's regional information is determined based on the first pronunciation information and the second pronunciation information corresponding to multiple regional accents. Candidate objects are recommended to the user based on the regional information. It is possible to infer the user's regional information from the accent in the user's voice data. Since the voice contains rich dialect or accent information, the user's regional information can be determined more accurately. In addition, the present application can take into account the situation of mixed accents and mixed regions through the second pronunciation information corresponding to multiple regional accents, and obtain more comprehensive regional information, thereby making the recommendation effect more comprehensive.

[0130] Figure 4 FIG. 2 shows a structural diagram of a personalized recommendation device according to an embodiment of the present application. Figure 4 As shown, the device comprises:

[0131] A first determination module 401 is used to obtain historical voice data of a user and obtain first pronunciation information of the user based on the historical voice data;

[0132] A second determination module 402, configured to determine the user's regional information based on the first pronunciation information and second pronunciation information corresponding to a plurality of regional accents;

[0133] The recommendation module 403 is used to recommend candidate objects to the user based on the user's region information.

[0134] In a possible implementation, the first determining module 401 is configured to:

[0135] Perform text transcription and pinyin transcription on the historical voice data to obtain a text sequence of the historical voice data and a pinyin syllable sequence pronounced by the user;

[0136] Determine a Mandarin pinyin sequence based on a text sequence;

[0137] Based on the association between the text sequence, the Mandarin Pinyin sequence and the Pinyin syllable sequence pronounced by the user, a triple corresponding to each Chinese character text in the text sequence is obtained.

[0138] In a possible implementation, the first determining module 401 is further configured to:

[0139] Among the triples corresponding to each Chinese character text in the text sequence, the triples whose Mandarin pronunciation is the same as the user's pronunciation are removed.

[0140] In a possible implementation, the second pronunciation information includes a first association relationship between a Chinese character text and a pronunciation in a corresponding regional accent, and a second association relationship between a Mandarin pronunciation and a pronunciation in a corresponding regional accent. The second determination module 402 is configured to:

[0141] The user's region information is determined based on the first pronunciation information and the first association relationship, and the first pronunciation information and the second association relationship.

[0142] In a possible implementation, the first pronunciation information includes a plurality of triples, each triple includes a Chinese character text, a Mandarin pronunciation, and a user pronunciation, and the user's regional information is determined according to the first pronunciation information and the first association relationship, and the first pronunciation information and the second association relationship, including:

[0143] For each triple in the plurality of triples, matching the Chinese text and the user pronunciation in the triple with the first association relationships corresponding to the plurality of regions respectively to obtain a first matching result;

[0144] For each triple in the plurality of triples, matching the Mandarin pronunciation and the user pronunciation in the triple with the second association relationships corresponding to the plurality of regions respectively to obtain a second matching result;

[0145] Based on the first matching result and the second matching result, the region information of the user is determined.

[0146] In a possible implementation, determining the user's region information based on the first matching result and the second matching result includes:

[0147] In response to the first matching result being a successful match, adding points to the corresponding regional tag;

[0148] In response to the second matching result being a successful match, adding points to the corresponding regional tag;

[0149] The regional tags and their scores are used as the user's regional information.

[0150] In a possible implementation, the device further includes:

[0151] A third determination module, configured to determine a plurality of region tags according to the region information of the pre-collected data;

[0152] The collection module is used to collect regional accent pronunciation libraries for multiple regional tags respectively, and each regional accent pronunciation library stores the second pronunciation information of the corresponding region.

[0153] In a possible implementation, the collection module is used to:

[0154] Collect pronunciation information of regional accents corresponding to each regional label that is different from Mandarin;

[0155] Based on the pronunciation information of the regional accent that is different from that of Mandarin, a key-value pair representing a first association relationship and a key-value pair representing a second association relationship are determined; the key of the key-value pair representing the first association relationship is a Chinese text, and the value is the pronunciation of the corresponding regional accent; the key of the key-value pair representing the second association relationship is a Mandarin pronunciation, and the value is the pronunciation of the corresponding regional accent;

[0156] A regional accent pronunciation library is determined based on the key-value pair representing the first association relationship and the key-value pair representing the second association relationship.

[0157] According to an embodiment of the present application, the user's historical voice data is obtained to determine the user's regional information, so that the user does not need to actively provide the regional information. The user's first pronunciation information is obtained based on the historical voice data, and the user's regional information is determined based on the first pronunciation information and the second pronunciation information corresponding to multiple regional accents. Candidate objects are recommended to the user based on the regional information. It is possible to infer the user's regional information from the accent in the user's voice data. Since the voice contains rich dialect or accent information, the user's regional information can be determined more accurately. In addition, the present application can take into account the situation of mixed accents and mixed regions through the second pronunciation information corresponding to multiple regional accents, and obtain more comprehensive regional information, thereby making the recommendation effect more comprehensive.

[0158] In some embodiments, the functions or modules included in the device provided by the embodiments of the present disclosure can be used to execute the method described in the above method embodiments. The specific implementation can refer to the description of the above method embodiments, and for the sake of brevity, it will not be repeated here.

[0159] The embodiment of the present disclosure also provides a computer-readable storage medium on which computer program instructions are stored, and the computer program instructions implement the above method when executed by a processor. The computer-readable storage medium can be a volatile or non-volatile computer-readable storage medium.

[0160] The embodiment of the present disclosure further proposes a personalized recommendation device, comprising: a processor; a memory for storing instructions executable by the processor; wherein the processor is configured to implement the above method when executing the instructions stored in the memory.

[0161] The embodiments of the present disclosure also provide a computer program product, including a computer-readable code, or a non-volatile computer-readable storage medium carrying the computer-readable code. When the computer-readable code runs in a processor of an electronic device, the processor in the electronic device executes the above method.

[0162] Figure 5 1 is a block diagram of a device 1900 for personalized recommendation according to an exemplary embodiment. For example, the device 1900 may be provided as a server or a terminal device. Figure 5 , the apparatus 1900 includes a processing component 1922, which further includes one or more processors, and a memory resource represented by a memory 1932 for storing instructions, such as an application, that can be executed by the processing component 1922. The application stored in the memory 1932 may include one or more modules, each corresponding to a set of instructions. In addition, the processing component 1922 is configured to execute instructions to perform the above method.

[0163] The device 1900 may also include a power supply component 1926 configured to perform power management of the device 1900, a wired or wireless network interface 1950 configured to connect the device 1900 to a network, and an input / output interface 1958 (I / O interface). The device 1900 may operate based on an operating system stored in the memory 1932, such as Windows Server 2000. TM , MacOS X TM , Unix TM ,Linux TM , FreeBSD TM or similar.

[0164] In an exemplary embodiment, a non-volatile computer-readable storage medium is also provided, such as a memory 1932 including computer program instructions, which can be executed by the processing component 1922 of the device 1900 to perform the above method.

[0165] The present disclosure may be a system, a method and / or a computer program product. The computer program product may include a computer-readable storage medium carrying computer-readable program instructions for causing a processor to implement various aspects of the present disclosure.

[0166] A computer-readable storage medium may be a tangible device that can hold and store instructions used by an instruction execution device. A computer-readable storage medium may be, for example, but not limited to, an electrical storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any suitable combination of the foregoing. More specific examples of computer-readable storage media (a non-exhaustive list) include: a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a static random access memory (SRAM), a portable compact disk read-only memory (CD-ROM), a digital versatile disk (DVD), a memory stick, a floppy disk, a mechanical encoding device, such as a punch card or a raised structure in a groove on which instructions are stored, and any suitable combination of the foregoing. As used herein, a computer-readable storage medium is not to be interpreted as a transient signal per se, such as a radio wave or other freely propagating electromagnetic wave, an electromagnetic wave propagating through a waveguide or other transmission medium (e.g., a light pulse through a fiber optic cable), or an electrical signal transmitted through a wire.

[0167] The computer-readable program instructions described herein can be downloaded from a computer-readable storage medium to each computing / processing device, or downloaded to an external computer or external storage device via a network, such as the Internet, a local area network, a wide area network, and / or a wireless network. The network can include copper transmission cables, optical fiber transmissions, wireless transmissions, routers, firewalls, switches, gateway computers, and / or edge servers. The network adapter card or network interface in each computing / processing device receives the computer-readable program instructions from the network and forwards the computer-readable program instructions for storage in the computer-readable storage medium in each computing / processing device.

[0168] The computer program instructions for performing the operation of the present disclosure may be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-related instructions, microcode, firmware instructions, state setting data, or source code or object code written in any combination of one or more programming languages, including object-oriented programming languages, such as Smalltalk, C++, etc., and conventional procedural programming languages, such as "C" language or similar programming languages. Computer-readable program instructions may be executed completely on a user's computer, partially on a user's computer, as an independent software package, partially on a user's computer, partially on a remote computer, or completely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., using an Internet service provider to connect via the Internet). In some embodiments, an electronic circuit, such as a programmable logic circuit, a field programmable gate array (FPGA), or a programmable logic array (PLA), may be customized by utilizing the state information of the computer-readable program instructions, and the electronic circuit may execute the computer-readable program instructions, thereby realizing various aspects of the present disclosure.

[0169] Various aspects of the present disclosure are described herein with reference to the flowcharts and / or block diagrams of the methods, devices (systems) and computer program products according to the embodiments of the present disclosure. It should be understood that each box in the flowchart and / or block diagram and the combination of each box in the flowchart and / or block diagram can be implemented by computer-readable program instructions.

[0170] These computer-readable program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, thereby producing a machine, so that when these instructions are executed by the processor of the computer or other programmable data processing device, a device that implements the functions / actions specified in one or more boxes in the flowchart and / or block diagram is generated. These computer-readable program instructions can also be stored in a computer-readable storage medium, and these instructions cause the computer, programmable data processing device, and / or other equipment to work in a specific manner, so that the computer-readable medium storing the instructions includes a manufactured product, which includes instructions for implementing various aspects of the functions / actions specified in one or more boxes in the flowchart and / or block diagram.

[0171] Computer-readable program instructions may also be loaded onto a computer, other programmable data processing apparatus, or other device so that a series of operating steps are performed on the computer, other programmable data processing apparatus, or other device to produce a computer-implemented process, thereby causing the instructions executed on the computer, other programmable data processing apparatus, or other device to implement the functions / actions specified in one or more boxes in the flowchart and / or block diagram.

[0172] The flow chart and block diagram in the accompanying drawings show the possible architecture, function and operation of the system, method and computer program product according to multiple embodiments of the present disclosure. In this regard, each square box in the flow chart or block diagram can represent a part of a module, program segment or instruction, and a part of the module, program segment or instruction includes one or more executable instructions for realizing the specified logical function. In some alternative implementations, the function marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two continuous square boxes can actually be executed substantially in parallel, and they can sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs the specified function or action, or can be implemented with a combination of special hardware and computer instructions.

[0173] The embodiments of the present disclosure have been described above, and the above description is exemplary, not exhaustive, and is not limited to the disclosed embodiments. Many modifications and changes will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the described embodiments. The selection of terms used herein is intended to best explain the principles of the embodiments, practical applications, or technical improvements in the market, or to enable other persons of ordinary skill in the art to understand the embodiments disclosed herein.

Claims

1. A personalized recommendation method, characterized in that: The method comprises: Acquire historical voice data of the user, and obtain first pronunciation information of the user based on the historical voice data; Determining the user's regional information based on the first pronunciation information and second pronunciation information corresponding to a plurality of regional accents; Recommend candidate objects to the user based on the user's region information.

2. The method according to claim 1, characterized in that: The obtaining the first pronunciation information of the user based on the historical voice data includes: Performing text transcription and pinyin transcription on the historical voice data to obtain a text sequence of the historical voice data and a pinyin syllable sequence pronounced by the user; Determine a Mandarin pinyin sequence based on the text sequence; Based on the association relationship between the text sequence, the Mandarin Pinyin sequence and the Pinyin syllable sequence pronounced by the user, a triple corresponding to each Chinese character text in the text sequence is obtained.

3. The method according to claim 2, characterized in that The obtaining of the first pronunciation information of the user based on the historical voice data further includes: Among the triples corresponding to each Chinese character text in the text sequence, the triples whose Mandarin pronunciation is the same as the user pronunciation are removed.

4. The method according to claim 1, characterized in that The second pronunciation information includes a first association relationship between a Chinese character text and a pronunciation in a corresponding regional accent, and a second association relationship between a Mandarin pronunciation and a pronunciation in a corresponding regional accent. The determining of the user's regional information according to the first pronunciation information and the second pronunciation information corresponding to the plurality of regional accents includes: The user's region information is determined based on the first pronunciation information and the first association relationship, and the first pronunciation information and the second association relationship.

5. The method according to claim 4, characterized in that The first pronunciation information includes a plurality of triples, each triple including a Chinese character text, a Mandarin pronunciation, and a user pronunciation, and determining the user's regional information according to the first pronunciation information and the first association relationship, and the first pronunciation information and the second association relationship, includes: For each triple in the plurality of triples, matching the Chinese text and the user pronunciation in the triple with the first association relationships corresponding to the plurality of regions respectively to obtain a first matching result; For each triple in the plurality of triples, matching the Mandarin pronunciation and the user pronunciation in the triple with the second association relationships corresponding to the plurality of regions respectively to obtain a second matching result; Based on the first matching result and the second matching result, the region information of the user is determined.

6. The method according to claim 5, characterized in that The determining the user's region information based on the first matching result and the second matching result includes: In response to the first matching result being a successful match, adding points to the corresponding regional tag; In response to the second matching result being a successful match, adding points to the corresponding regional tag; The regional tags and their scores are used as the user's regional information.

7. The method according to claim 5, characterized in that The method further comprises: Determine multiple regional tags based on regional information of pre-collected data; Regional accent pronunciation libraries are collected for each of the multiple regional tags, and each regional accent pronunciation library stores second pronunciation information of the corresponding region.

8. The method according to claim 7, characterized in that The method of collecting regional accent pronunciation libraries for multiple regional tags respectively includes: Collect pronunciation information of regional accents corresponding to each regional label that is different from Mandarin; Based on the pronunciation information of the regional accent that is different from that of Mandarin, a key-value pair representing the first association relationship and a key-value pair representing the second association relationship are determined; the key in the key-value pair representing the first association relationship is a Chinese text, and the value is a pronunciation in the corresponding regional accent; the key in the key-value pair representing the second association relationship is a Mandarin pronunciation, and the value is a pronunciation in the corresponding regional accent; A regional accent pronunciation library is determined based on the key-value pair representing the first association relationship and the key-value pair representing the second association relationship.

9. A personalized recommendation device, characterized in that: The device comprises: A first determination module, configured to obtain historical voice data of a user, and obtain first pronunciation information of the user based on the historical voice data; A second determination module, configured to determine the user's regional information based on the first pronunciation information and second pronunciation information corresponding to a plurality of regional accents; The recommendation module is used to recommend candidate objects to the user based on the user's regional information.

10. A personalized recommendation device, characterized in that: include: processor; a memory for storing processor-executable instructions; The processor is configured to implement the method described in any one of claims 1 to 8 when executing the instructions stored in the memory.

11. A non-volatile computer-readable storage medium having computer program instructions stored thereon, characterized in that: When the computer program instructions are executed by a processor, the method according to any one of claims 1 to 8 is implemented.

Images