Intelligent recommendation method and device based on mixed modal opinion consensus decision, and medium

Abstract

The invention discloses an intelligent recommendation method and device based on a mixed modal opinion consensus decision, and a medium. The method comprises the following steps: collecting numerical opinions and language opinions of a plurality of users for a target question to form a mixed opinion set; searching a user trust set of each user, and calculating trust weights among the users; converting the numerical value opinions into language opinions and converting the language opinions into numerical value opinions; the converted numerical value opinions and language opinions are evolved; and performing mixed opinion conversion, trust weight calculation and mixed opinion evolution through repeated iteration until a consensus is achieved, and determining recommendation result output according to a consensus state. By establishing a mutual conversion mechanism of numerical value opinions and language opinions of users, mutual conversion between opinions is realized, numerical value opinions and language opinions can be processed at the same time, and an efficient and flexible mixed opinion evolution framework is established, so that an accurate personalized recommendation result is finally generated through group consensus, and efficient and accurate intelligent recommendation is realized.

Description

Technical Field

[0001] The present invention relates to the field of intelligent recommendation technology, and in particular to an intelligent recommendation method, device and medium based on hybrid modal opinion consensus decision-making. Background Art

[0002] In the current digital economy era, users are faced with a vast number of choices of goods and services. By using data analysis, machine learning algorithms and other methods to predict users' ratings of goods or services, and thus accurately recommend goods or services to users, intelligent recommendations of goods or services can be achieved, which can greatly improve the user experience. Intelligent recommendations require the analysis of a large number of users' historical evaluation data to filter out the highest-rated goods or services. However, users may express their opinions and rate goods or services in different ways. For example, some users may express their opinions through numerical ratings (such as 0 to 10 points), while some users may use language tags (such as "good reviews", "medium reviews", "bad reviews", etc.). It is difficult to directly compare or merge the expression formats of opinions in different modes, which makes it difficult to accurately determine user preferences based on users' different evaluation modes of goods or services.

[0003] The evolution of group opinions and consensus decision-making is that in social networks or group interactions, individual opinions gradually form group consensus through mutual influence and adjustment, and finally reach a consensus through group interaction and opinion adjustment. Based on the group opinion evolution model, the recommendation results can be optimized by analyzing the opinion differences and interactive relationships between users. Based on the consensus decision-making method, by simulating group interaction and opinion adjustment, more accurate recommendation results can be provided for the intelligent recommendation system. In the prior art, the group opinion evolution model usually adopts a limited confidence model, but the traditional limited confidence model assumes that individuals will only interact when their opinion gap is less than a certain trust threshold. Therefore, it can only be applied to opinions in a single expression format and cannot handle heterogeneous mixed-modal opinion expression formats (numerical and linguistic). The mixed expression of numerical and linguistic opinions cannot be effectively converted and integrated within the framework of the trust mechanism, making it impossible to fully utilize the group consensus decision-making mechanism to accurately capture user preferences and ensure the accuracy and reliability of recommendations. Summary of the invention

[0004] The technical problem to be solved by the present invention is: in response to the technical problems existing in the prior art, the present invention provides an intelligent recommendation method, device and medium based on hybrid modal opinion consensus decision-making, which has a simple implementation method, low cost, high accuracy and reliability, and can fully integrate the user's numerical and linguistic hybrid modal opinions to reach a consensus and realize accurate intelligent recommendation.

[0005] In order to solve the above technical problems, the technical solution proposed by the present invention is: An intelligent recommendation method based on hybrid modal opinion consensus decision-making, comprising the following steps: Collecting multiple users' numerical opinions and language opinions on the target question to form a mixed opinion set; the numerical opinion is a numerical value representing the degree of the user's emotional tendency towards the target question, and the language opinion is a text representing the type of the user's emotional tendency towards the target question; Mixed opinion conversion: converting the numerical opinions in the mixed opinion set into language opinions and converting the language opinions into numerical opinions, respectively obtaining a converted numerical opinion set and a converted language opinion set; when converting the numerical opinions into language opinions, converting the numerical opinions expressed by other users into language opinions to form a unified language opinion format; when converting the language opinions into numerical opinions, converting the language opinions expressed by other users into numerical opinions to form a unified numerical opinion format; Trust weight calculation: Search the user trust set of each user based on the current mixed opinion set, where the user trust set expressing numerical opinions contains numerical opinions whose absolute values ​​of the differences of the numerical opinions of users at different times are within the numerical trust threshold range, and the user trust set expressing language opinions contains language opinions whose absolute values ​​of the differences of the language opinions of users at different times are within the language trust threshold range, and calculate the trust weights between users based on the user trust sets of each user; Mixed opinion evolution: the converted numerical opinion set and language opinion set are evolved. During the evolution process, the numerical opinions and language opinions at the previous moment and the trust weights between the users are used to calculate the numerical opinions and language opinions at the next moment. After the evolution is completed, a new mixed opinion set is formed; The mixed opinion conversion, trust weight calculation and mixed opinion evolution are iteratively performed until the language opinions and numerical opinions in the iterative mixed opinion set reach a consensus, and the recommended result output is determined according to the current consensus state.

[0006] Furthermore, the numerical opinions Convert to language bigram The expression is:

[0007] in, It represents the deviation between the user's converted language opinion and numerical opinion. , Expressing opinions about values The function of Expressing numerical opinion The converted language tuples are Indicates the user's language opinion. represents the approximate rounding function, Indicates the user's language granularity, ; Language bigrams Convert to numerical opinion The expression is:

[0008] in, Represents the inverse function of the opinion about the value.

[0009] Further, searching for the user trust set of each user according to the current mixed opinion set includes: User corresponding to the numerical opinion According to yourself t Numerical opinion and numerical confidence threshold at the moment Find out t User trust set at the moment :

[0010] in, Indicates to users All users found in t The set of numerical opinions at time, Indicates the found user. Indicates user exist t The numerical opinion at the moment, Indicates user exist t The numerical opinion at the moment, where ; represents the numerical confidence threshold, ; User corresponding to language opinion According to yourself t Language opinion and language trust threshold at the moment Find out t User trust set at the moment :

[0011] in, Indicates to users All users found in t The collection of language opinions at the moment, Indicates user exist t The language opinions of the moment, Indicates user exist t The language opinions of the moment, , represents the language trust threshold, , Indicates the language granularity of the user.

[0012] Furthermore, in calculating the trust weights between users according to the user trust sets of each user, the user trust weights are calculated according to the following formula: For users Trust weight :

[0013] in, Indicates user exist t The trust set of other users in their opinions when expressing mixed opinions at any time The number of elements in Indicates user The self-confidence value, Indicates that the user is t A mixed set of opinions at the moment, Trust Set Total number of users in the system.

[0014] Furthermore, evolving the converted numerical opinion set and language opinion set includes: The converted numerical opinions are evolved using a numerical opinion evolution model under limited confidence, and the numerical opinion evolution model is:

[0015] in, , Represents the numerical opinion user exist , t The numerical opinion at the moment, Represents the numerical opinion user exist t The numerical opinion at the moment, Indicates numerical opinion user For users The trust weight of Indicates numerical opinion user The self-confidence value, Indicates numerical opinion user exist t The total number of users in the trust set of other users’ opinions when expressing numerical opinions at a moment; The converted numerical opinions are evolved using a language opinion evolution model under limited confidence, and the language opinion evolution model is:

[0016] in, , User who expresses language opinion exist , t The language opinions of the moment, Indicates user exist t The language opinions of the moment, User who expresses language opinion For users The trust weight of User who expresses language opinion The self-confidence value, User who expresses language opinion exist t The total number of users in the trust set of other users' opinions when expressing their opinions at a certain moment.

[0017] Further, the iterative execution of the mixed opinion conversion, trust weight calculation and mixed opinion evolution until the language opinions and numerical opinions in the iterative mixed opinion set reach a consensus includes: Get the mixed opinion set obtained after each iteration; If the obtained mixed opinion set is converted into a numerical opinion set that satisfies the opinion set ,and And converted into a language opinion set to satisfy the opinion set ,and ,in ~ Indicates various numerical opinions, ~ Express opinions in various languages, It represents the number of numerical opinions or language opinions, and if the deviation of the iteration result value is within the preset allowable range, it is determined that a consensus is reached.

[0018] An intelligent recommendation device based on hybrid modal opinion consensus decision-making, comprising: A mixed opinion collection module is used to collect numerical opinions and language opinions of multiple users on the target problem to form a mixed opinion set; the numerical opinion is a numerical value used to represent the degree of the user's emotional tendency towards the target problem, and the language opinion is a text-based representation of the user's emotional tendency type towards the target problem; A mixed opinion conversion module, used to convert the numerical opinions in the mixed opinion set into language opinions and convert the language opinions into numerical opinions, so as to obtain converted numerical opinions and language opinions; A trust weight calculation module is used to search for a user trust set of each user based on the current mixed opinion set, wherein the user trust set expressing numerical opinions includes numerical opinions whose absolute values ​​of differences between numerical opinions of users at different times are within a numerical trust threshold range, and the user trust set expressing linguistic opinions includes linguistic opinions whose absolute values ​​of differences between linguistic opinions of users at different times are within a linguistic trust threshold range, and the trust weights between users are calculated based on the user trust sets of each user; A mixed opinion evolution module, used to evolve the converted numerical opinion set and language opinion set, during which the numerical opinions and language opinions at the previous moment and the trust weights between the users are used to calculate the numerical opinions and language opinions at the next moment, and a new mixed opinion set is formed after the evolution is completed; The iterative output module is used to iteratively execute the mixed opinion transformation, trust weight calculation and mixed opinion evolution until the language opinions and numerical opinions in the iterative mixed opinion set reach a consensus, and determine the recommended result output according to the current consensus state.

[0019] An electronic device comprises a processor and a memory, wherein the memory is used to store a computer program, and the processor is used to execute the computer program to perform the above method.

[0020] A computer-readable storage medium storing a computer program, wherein the computer program implements the above method when executed by a processor.

[0021] Compared with the prior art, the advantages of the present invention are: the present invention realizes the mutual conversion between numerical opinions and language opinions by establishing a mutual conversion mechanism between user numerical opinions and language opinions, calculates the trust weights between users by using user trust sets, and evolves the mixed opinion sets based on the calculated trust weights to form new mixed opinion sets, and repeatedly iterates and updates the trust sets and the opinion evolution process until the group opinions reach a consensus, thereby determining the intelligent recommendation results. The present invention can process numerical opinions and language opinions at the same time, establish an efficient and flexible mixed opinion evolution framework, effectively deal with the disagreement between numerical opinions and language opinions, and enable individuals with different expression formats to communicate effectively. At the same time, based on the consensus mechanism, the group is helped to form a consistent decision under multimodal opinion expression, thereby ultimately generating accurate personalized recommendation results through group consensus, achieving efficient and accurate intelligent recommendations, and improving user experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 It is a schematic diagram of the implementation flow of the intelligent recommendation method based on hybrid modal opinion consensus decision-making in this embodiment. DETAILED DESCRIPTION

[0023] The present invention is further described below in conjunction with the accompanying drawings and specific preferred embodiments, but the protection scope of the present invention is not limited thereby.

[0024] Take the intelligent e-commerce recommendation system as an example. The intelligent e-commerce recommendation system can help users find goods or services of interest. In the existing technology, such recommendation systems usually rely only on users' numerical ratings (such as star ratings, purchase volume, browsing time, etc.), or only rely on language tags in user comments (such as "good reviews", "medium reviews", "bad reviews", etc.). However, in practice, numerical ratings and language tags are often intertwined. A single data expression method cannot fully reflect the complex preferences of users. Especially in a diverse user group, the differences in the way individuals express their opinions may be large. The recommendation results directly determined based on a single type of evaluation data are not accurate. In addition, individuals expressing discrete language opinions are prone to fuzzy results and lack of precision, and continuous numerical opinion expressions are also uncertain. Both single numerical opinions and language opinions lack certainty, and there are differences between discrete language preferences and continuous numerical preferences, which makes it difficult to directly establish a corresponding opinion conversion mechanism between the two and a fusion evolution mechanism for mixed opinion expressions.

[0025] The present invention first collects users' numerical opinions and language opinions on a certain product or service issue to form an initial mixed opinion set, and at the same time establishes a mutual conversion mechanism between users' numerical opinions and language opinions to achieve mutual conversion between numerical opinions and language opinions, then finds out the user trust set to calculate the trust weights between users, and based on the calculated trust weights, respectively evolves the numerical opinions and language opinions in the mixed opinion set to form a new mixed opinion set, and iterates and updates the trust set and opinion evolution process until the group opinion reaches a consensus, thereby determining the recommendation result for the product or service. The present invention can process numerical opinions and language opinions at the same time, establish an efficient and flexible mixed opinion evolution framework, effectively deal with the disagreement between numerical opinions and language opinions, and enable individuals with different expression formats to communicate effectively. At the same time, based on the consensus mechanism, the group is helped to form a consistent decision under multimodal opinion expression, so as to finally generate accurate personalized recommendation results through group consensus, realize efficient and accurate intelligent recommendation, and improve user experience.

[0026] Taking the intelligent e-commerce recommendation system as an example, the evaluation details of the e-commerce platform contain a large number of users' opinions on the products in different ways: one is the numerical rating, such as giving a certain electronic product a rating of 4 stars; the other is to provide language comments, such as: "I like this mobile phone very much, the performance is very good, I recommend buying it!" and so on. By adopting the method of the present invention, the numerical ratings and language comments of all users can be transformed and integrated to form a comprehensive group opinion. If the language opinion and the numerical opinion finally reach a consensus, the consensus decision result will be used as the corresponding recommendation result. For example, if the group rating of a certain product is 4.5 stars, and the consensus reached in the language comments is "I like it very much", then the product will be recommended as a high priority; correspondingly, if the group opinion of a certain product is negative after reaching a consensus, the recommendation of the product will be reduced accordingly.

[0027] like Figure 1 As shown, the detailed steps of the intelligent recommendation method based on hybrid modal opinion consensus decision in this embodiment include: Step S01. Collect numerical opinions and language opinions of multiple users' target questions to form a mixed opinion set; the numerical opinions are numerical values ​​used to represent the degree of the user's emotional tendency towards the target question, and the language opinions are texts used to represent the type of the user's emotional tendency towards the target question.

[0028] This embodiment first defines the range of numerical opinions and language opinions, collects the opinions of the same group of users on issues such as a product or service, forms an initial mixed opinion set, and sets a self-trust value for the user.

[0029] Specifically, you can define a numerical opinion score of 0-10 points (configurable), a language opinion range ("good", "medium", "bad"), that is, , the language granularity is . Collect the same target user group The numerical opinion scores and linguistic opinions of , uniformly stored in the database, ensuring that the data format is uniform and complete, and indicating the type of each user's feedback (numerical or linguistic), while cleaning to remove invalid or incomplete feedback, and organizing to form the user's initial mixed opinion set ,in, Indicates user The numerical opinion at the moment, Indicates user Language opinion at the moment; set user Self-confidence value Specifically, you can set the self-trust value of all users to be the same .

[0030] It is understandable that both the numerical opinion score and the language opinion range can be configured according to actual needs. For example, the language opinion range can also be configured as "agree", "neutral", "disagree", etc.

[0031] Step S02. Mixed opinion conversion: converting the numerical opinions in the mixed opinion set into language opinions and converting the language opinions into numerical opinions, respectively obtaining the converted numerical opinion set and language opinion set; when converting the numerical opinions into language opinions, converting the numerical opinions expressed by other users into language opinions to form a unified language opinion format; when converting the language opinions into numerical opinions, converting the language opinions expressed by other users into numerical opinions to form a unified numerical opinion format; In this embodiment, converting language opinions into numerical opinions means that users who express numerical opinions can only merge numerical opinions, and therefore it is necessary to convert language opinions expressed by other users into numerical opinions to form a unified numerical opinion format, so that the numerical opinions of each user can be merged for opinion update later; converting numerical opinions into language opinions means that users who express language opinions can only merge language opinions, and therefore it is necessary to convert numerical opinions expressed by other users into language opinions to form a unified language opinion format, so that the language opinions of each user can be merged for opinion update later.

[0032] This embodiment establishes a mutual conversion mechanism between user numerical opinions and language opinions to achieve mutual conversion between numerical and language opinions, converting numerical opinions into language opinions and also converting language opinions into numerical opinions, so that the two can be compared and integrated under a unified framework, thereby enabling individuals with different opinion expression formats to better communicate with each other. For example, language tags such as "agree" can be mapped to numerical intervals, and numerical ratings can also be converted into language tags.

[0033] Specifically, we can introduce language granularity based on the bigram model Construct a first conversion model that converts numerical opinions into language opinions. For example, the following first conversion model can be used to convert numerical opinion data Convert to language bigrams , The initial value of is 0, that is, the language tuple : (1) in, It represents the deviation between the user's converted language opinion and numerical opinion. , Express opinions about values The function of Expressing numerical opinion The converted language tuples are Indicates the user's language opinion. represents the approximate rounding function, Indicates the user's language granularity, .

[0034] The bigram model is a mathematical model that handles uncertainty and ambiguity by representing fuzzy information or fuzzy language as a bigram. Each bigram consists of a linguistic term and a numerical value, and the numerical value range is between (-0.5, 0.5). This model allows continuous representation in the domain of linguistic information, so that any information can be represented in the aggregation process. This embodiment constructs the above-mentioned first conversion model based on the bigram model, mapping the high precision of numerical opinions (floating decimals between 0 and 10) to the discretized form of linguistic opinions, which can better quantify and integrate the conversion of individual numerical opinions to linguistic opinions.

[0035] As shown in the first transformation model of formula (1), by using Corresponding language comments , numerical opinions can be mapped to discrete language tags, so that fuzzy numerical information can be matched with predefined language tags, and the mapping process can be simplified by rounding, ignoring the high-precision details of unimportant numerical opinions, avoiding excessive refinement of numerical values, and paying more attention to the relative strength of these opinions, so that it is easy to quickly map to a coarser language category. At the same time, the rounding operation can also effectively map the numerical opinions to a fixed language tag interval. For example, the value 2.5 can be rounded to 3, which can be mapped to the "neutral" category. After the rounding operation, the match between the numerical value and the language tag is more intuitive and clear. At the same time, by introducing the user's language granularity Normalizing the numerical opinion values ​​can effectively map the numerical opinions to standardized values ​​on the [0,1] interval.

[0036] Furthermore, in order to convert the language labels in the language binary back to the corresponding numerical intervals and determine the specific numerical scores to complete the reverse mapping, a second conversion model for converting language opinions into numerical opinions can be constructed based on the inverse function of the conversion model for converting numerical opinions into language opinions, so as to realize the conversion of language opinions into numerical opinions, that is, the process of converting language binary into numerical opinions is the inverse function of converting numerical opinions into language binary. For example, based on fuzzy set theory and the principle of membership (numerical) function, the following second conversion model can be used to convert language binary into numerical opinions. Convert to numerical opinion : (2) in, Represents the inverse function of the opinion about the value.

[0037] According to the above formula, the subscript of the language opinion can be mapped to a specific value for calculation, where When the language opinion is converted into a numerical opinion for the first time, it is 0. After the first evolution, a new , that is, according to right Perform calculations.

[0038] Traditional bigram models may encounter inconsistency or incomparability when dealing with standardization between different language tags and comparison between numerical values. This embodiment first constructs a first conversion model for converting numerical opinions into language opinions based on the bigram model, and then constructs a second conversion model for converting language opinions into numerical opinions based on the inverse function of the first conversion model. The numerical opinions are restored or inferred using language tags and membership degrees, and the language and numerical opinions can be standardized and normalized, and the scales of numerical opinions and language opinions are unified, so that the two can be compared and converted under the same framework, thereby adapting to the complexity of multimodal opinions.

[0039] This embodiment, by adopting the above-mentioned mutual conversion mechanism between user numerical opinions and language opinions, can break the limitation of traditional use of only single numerical opinions or language opinions, so that different types of opinions can be converted into each other under an appropriate mechanism, thereby efficiently handling the evolution of mixed opinions among groups.

[0040] Step S03. Trust weight calculation: Search the user trust set of each user based on the current mixed opinion set, where the user trust set expressing numerical opinions includes numerical opinions whose absolute values ​​of the differences between the numerical opinions of users at different times are within the numerical trust threshold range, and the user trust set expressing language opinions includes language opinions whose absolute values ​​of the differences between the language opinions of users at different times are within the language trust threshold range. The trust weights between users are calculated based on the user trust sets of each user.

[0041] This embodiment sets a language trust threshold, finds out the user trust set, and then calculates the user's trust weight for other users.

[0042] Specifically, searching for the user trust set of each user according to the current mixed opinion set includes: User corresponding to the numerical opinion According to yourself t Numerical opinion and numerical confidence threshold at the moment Find out t User trust set at the moment : (3) in, Indicates to users All users found in t The set of numerical opinions at time, Indicates the found user. Indicates user exist t The numerical opinion at the moment, Indicates user exist t The numerical opinion at the moment, where ; represents the numerical confidence threshold, .

[0043] User corresponding to language opinion According to yourself t Language opinion and language trust threshold at the moment Find out t User trust set at the moment : (4) in, Indicates to users All users found in t The collection of language opinions at the moment, Indicates user exist t The language opinions of the moment, Indicates user exist t The language opinions of the moment, , represents the language trust threshold, , Indicates the language granularity of the user.

[0044] As an optional implementation, in calculating the trust weights between users according to the user trust sets of each user, the trust weight distribution method of the HK model can be used to calculate the trust weights of the users according to the following formula: For users Trust weight : (5) in, Indicates user exist t The trust set of other users in their opinions when expressing mixed opinions at any time The number of elements in Indicates user The self-confidence value, Indicates that the user is t A mixed set of opinions at the moment, Trust Set Total number of users, users is a user in the mixed opinion set. The above trust weight calculation method is applicable to users corresponding to numerical opinions and users corresponding to language opinions.

[0045] When the same group of users express different opinions on the same event, each user has a certain self-trust value for himself and a certain trust value for other users except his own opinions. This embodiment constructs a trust weight calculation model based on the trust weight distribution method of the HK model. This calculation model can be used to evenly distribute the trust weights of other users' opinions except for the self-trust value of the user's own opinion, and effectively calculate the trust weight value of each user for other users.

[0046] Step S04. Mixed opinion evolution: Evolve the converted numerical opinion set and language opinion set. During the evolution process, the numerical opinions and language opinions at the previous moment and the trust weights between users are used to calculate the numerical opinions and language opinions at the next moment. After the evolution is completed, a new mixed opinion set is formed.

[0047] This embodiment forms a new mixed opinion set by respectively constructing a numerical opinion evolution model under limited confidence and a language opinion evolution model under limited confidence, evolving the opinions of the numerical opinion expressers according to the numerical opinion evolution model, and evolving the opinions of the language opinion expressers according to the language opinion evolution model.

[0048] Specifically, evolving the converted numerical opinions and language opinions includes: The converted numerical opinions are evolved using a numerical opinion evolution model under limited confidence, and the numerical opinion evolution model is: (6) in, , Represents the numerical opinion user exist , t The numerical opinion at the moment, Represents the numerical opinion user exist t The numerical opinion at the moment, Indicates numerical opinion user For users The trust weight of Indicates numerical opinion user The self-confidence value, Indicates numerical opinion user exist t The total number of users in the trust set of other users’ opinions when expressing numerical opinions at a moment; The converted numerical opinions are evolved using a language opinion evolution model under limited confidence, and the language opinion evolution model is: (7) in, , User who expresses language opinion exist , t The language opinions of the moment, Indicates user exist t The language opinions of the moment, User who expresses language opinion For users The trust weight of User who expresses language opinion The self-confidence value, User who expresses language opinion exist t The total number of users in the trust set of other users' opinions when expressing their opinions at a certain moment.

[0049] This embodiment constructs the numerical opinion evolution model under limited confidence (Formula (6)) to convert language opinions into numerical opinions before evolution, and constructs the language opinion evolution model under limited confidence (Formula (7)) to convert numerical opinions into language opinions before evolution. After each evolution is completed, The numerical opinions and language opinions at each moment are combined to form a new mixed opinion set and stored in the new mixed opinion set. This can evolve the different opinion formats of users and simulate the opinion evolution process under different trust levels, so that the model can adapt to the complex social network environment.

[0050] Step S05. Iteratively execute step S02 of mixed opinion conversion, step S03 of trust weight calculation, and step S04 of mixed opinion evolution until the language opinions and numerical opinions in the iterative mixed opinion set reach a consensus, and determine the recommended result output according to the current consensus status.

[0051] This embodiment identifies the user trust set corresponding to the new mixed opinion set through continuous iteration, calculates the trust weights between users, transforms the mixed opinions according to step S02, and evolves the mixed opinions according to step S04. If the mixed opinions reach a consensus, the recommended result output is determined according to the current consensus status.

[0052] Specifically, iteratively performing mixed opinion transformation, trust weight calculation, and mixed opinion evolution until the language opinions and numerical opinions in the iterative mixed opinion set reach a consensus includes: Get the mixed opinion set obtained after each iteration; If the obtained mixed opinion set is converted into a numerical opinion set that satisfies the opinion set ,and And converted into a language opinion set to satisfy the opinion set ,and ,in ~ Indicates various numerical opinions, ~ Express opinions in various languages, It represents the number of numerical opinions or language opinions, and if the deviation of the iteration result value is within the preset allowable range, it is determined that a consensus is reached.

[0053] This embodiment continuously iterates by utilizing the group opinion consensus mechanism. As more users participate in rating and commenting, the system can update the group opinion in real time and further optimize the recommendation results. If the original evaluation of a product or service is relatively divided, the recommendation strategy will be quickly adjusted after the above steps, so that real-time accurate and reliable intelligent recommendations can be achieved.

[0054] Through the above method, this embodiment can effectively analyze and convert the differences between numerical opinions and language opinions, help guide the group to achieve unity of group opinion when facing differences of opinion, and effectively solve the problem that traditional numerical ratings and language comments cannot be directly integrated, so that different types of user feedback can be effectively integrated to generate more accurate and reliable recommendation results.

[0055] The present invention can be applied to the intelligent recommendation of goods or services using e-commerce evaluation data, and can also be applied to the realization of group opinion consensus decision-making in fields with mixed opinion formats such as social media analysis. By adopting the above method of the present invention, the individual numerical opinions and language opinions are mutually transformed and evolved, which can promote the interaction of individuals with different expression formats, and then promote the formation of group opinion fusion consensus.

[0056] The present invention is further described below by taking the implementation of intelligent recommendation of a certain product by the present invention in a specific application embodiment as an example, and the detailed steps are as follows: Step S1: define the range of numerical opinions and language opinions, collect the evaluation opinions of the same group of users on a certain product, form an initial mixed opinion set, and set a self-trust value for the user.

[0057] Specifically, we define the numerical opinion score (0-10 points) and the language opinion range ("good review", "medium review", "bad review"), that is, , then the language granularity is . Get the same target user group The numerical opinion scores and language opinions are uniformly stored in the database, and the group size is , ensure that the data format is unified and complete and indicate the feedback type (numerical or verbal) of each user, perform data cleaning to remove invalid or incomplete feedback, and organize the initial mixed opinion set of users Among them, 0.4984, 0.5853, and 0.2238 are user The numerical opinions at the moment, and the rest are users The language opinion at the moment, at this time, set the self-trust value of all users , then the self-trust value matrix is: .

[0058] Step S2: convert the numerical opinions in the mixed opinion set into language opinions and convert the language opinions into numerical opinions to obtain converted numerical opinions and language opinions.

[0059] Specifically, the numerical opinion according to Convert to language bigram , the language bigram according to Convert to numerical opinion , according to mixed opinion The numerical opinions and language opinions are converted into: (1) For the language opinion expresser, the numerical opinion is converted into a language opinion, that is: The numerical opinion corresponding to 0.4984 is converted into language opinion:

[0060] The numerical opinion corresponding to 0.5853 is converted into language opinion:

[0061] The numerical opinion corresponding to 0.2238 is converted into language opinion:

[0062] at this time, , ; , ; , .

[0063] again ,The deviation caused by converting numerical opinions into language opinions is acceptable, which further verifies the rationality of the converted language opinions.

[0064] Therefore, in The language opinion after transformation is .

[0065] (2) For those who express numerical opinions, convert the linguistic opinions into numerical opinions, that is: 2 The corresponding linguistic opinions are converted into numerical opinions: , The initial value is 0; 1The corresponding linguistic opinion is converted into a numerical opinion: ; Therefore, in The numerical opinion after the moment transformation is .

[0066] Step S3: Set the numerical and language trust thresholds respectively, find out the user trust set, and calculate the trust weights between users.

[0067] Setting a numerical confidence threshold , and calculate the language trust threshold , then we can calculate .

[0068] Then find the user trust set, the numerical opinion expresser according to his numerical opinion and numerical trust threshold at time t Find the trust set at time t , the language opinion expresser is based on his language opinion and language trust threshold at time t Find the trust set at time t .

[0069] S33. Calculation of users (Numerical opinion expresser or verbal opinion expresser) to other users Trust weight .

[0070] (1) For the person expressing language opinions, the language opinion boundary trust value is ,but: Find out the user trust set of 2 corresponding language opinion holders: , the trust weights of other users are distributed as follows: ; Find the user trust set corresponding to the language opinion: , the trust weights of other users are distributed as follows: .

[0071] (2) For the person expressing numerical opinions, the boundary trust value of numerical opinions is ,but: Find the user trust set corresponding to 0.4984: , other individual trust weights are distributed as follows: ; Find the user trust set corresponding to 0.5853: , other individual trust weights are distributed as follows: ; Find the user trust set corresponding to 0.2238: , other individual trust weights are distributed as follows: .

[0072] S4. Use the numerical opinion evolution model under limited confidence and the language opinion evolution model under limited confidence to evolve the opinions of the numerical opinion expresser according to the numerical opinion evolution model, and evolve the opinions of the language opinion expresser according to the language opinion evolution model to form a new mixed opinion set.

[0073] First convert the language opinions into numerical opinions and then evolve them: , convert numerical opinions into language opinions and then evolve them: , respectively according to the numerical opinion evolution model and the language opinion evolution model in The opinion value obtained by evolving at all times is: 0.4984 corresponds to the following opinion: ; 2 Corresponding comments: ; 1 Corresponding comments: ; 0.5853 corresponds to the following opinion: ; 0.2238 corresponds to the following opinion: ; exist The new mixed opinion set is formed at this moment .

[0074] S5. Iteratively perform mixed opinion transformation, trust weight calculation, and mixed opinion evolution until the language opinions and numerical opinions in the mixed opinion set obtained by the iteration reach a consensus, and determine the recommended result output according to the current consensus status.

[0075] According to the above language opinion evolution steps, the language opinion undergoes the following 5 rounds of iterations to obtain the language opinion evolution results as shown in Table 1. Always be unified and reach consensus.

[0076] Table 1: Language opinion evolution process

[0077] Then, according to the above numerical opinion evolution steps, the numerical opinions undergo the following 5 rounds of iterations to obtain the numerical opinion evolution results as shown in Table 2. The group numerical opinions are The opinion values ​​at each moment are rounded to approximately the same value (the values ​​here allow rounding because the numerical opinions evolve and iterate over time, and the final opinion values ​​will be approximately the same).

[0078] Table 2: Numerical opinion evolution process

[0079] Finally, the mixed opinion dynamics leads to five iterations, as shown in Table 3.

[0080] Table 3: Mixed opinion iteration process

[0081] From Table 3, we can see that the user’s language opinions are A consensus was reached at the moment, and the numerical opinion was approximately 0.65, also Consensus was reached at all times, indicating that users can gradually form consensus opinions in the process of continuous iteration, and then determine the recommendation results based on the consensus decision results. At that moment, it can be converted into language opinion S1, then the opinions of all users in the mixed opinions have reached a consensus of S1, S1 corresponds to a medium rating, and the value also tends to be medium, so the group evaluation of the product is determined to be a medium rating, and it is configured as a medium recommendation level when making recommendations. From the above, it can be seen that the present invention can fully integrate numerical opinions and language opinions for communication and integration, and use group opinion consensus decision-making to ultimately achieve accurate and reliable intelligent recommendations.

[0082] This embodiment provides an intelligent recommendation device based on hybrid modal opinion consensus decision-making, including: A mixed opinion collection module is used to collect numerical opinions and language opinions of multiple users on the target problem to form a mixed opinion set; the numerical opinion is a numerical value used to represent the degree of the user's emotional tendency towards the target problem, and the language opinion is a text-based representation of the user's emotional tendency type towards the target problem; A mixed opinion conversion module, used to convert the numerical opinions in the mixed opinion set into language opinions and convert the language opinions into numerical opinions, so as to obtain converted numerical opinions and language opinions; A trust weight calculation module is used to search for a user trust set of each user according to the current mixed opinion set, the numerical opinions of the users whose absolute values ​​of the differences of the numerical opinions at different times are within the numerical trust threshold range, and the language opinions of the users whose absolute values ​​of the differences of the language opinions at different times are within the language trust threshold range, and calculate the trust weights between the users according to the user trust sets of each user; A mixed opinion evolution module, used to evolve the converted numerical opinions and language opinions. During the evolution process, the numerical opinions and language opinions at the previous moment and the trust weights between the users are used to calculate the numerical opinions and language opinions at the next moment. After the evolution is completed, a new mixed opinion set is formed; The iterative output module is used to iteratively execute the mixed opinion transformation, trust weight calculation and mixed opinion evolution until the language opinions and numerical opinions in the iterative mixed opinion set reach a consensus, and determine the recommended result output according to the current consensus state.

[0083] The intelligent recommendation device based on hybrid modal opinion consensus decision-making in this embodiment corresponds one to one with the above-mentioned intelligent recommendation method based on hybrid modal opinion consensus decision-making, and will not be described one by one here.

[0084] This embodiment further provides an electronic device, including a processor and a memory, wherein the memory is used to store a computer program, and the processor is used to execute the computer program to perform the above method.

[0085] It is understandable that the above method of this embodiment can be executed by a single device, such as a computer or server, etc., and can also be applied to a distributed scenario and completed by multiple devices in cooperation with each other. In the case of a distributed scenario, one of the multiple devices can only execute one or more steps in the above method of this embodiment, and multiple devices interact to complete the above method. The processor can be implemented in the form of a general-purpose CPU, a microprocessor, an application-specific integrated circuit, or one or more integrated circuits, etc., for executing related programs to implement the above method of this embodiment. The memory can be implemented in the form of a read-only memory ROM, a random access memory RAM, a static storage device, and a dynamic storage device. The memory can store an operating system and other applications. When the above method of this embodiment is implemented by software or firmware, the relevant program code is stored in the memory and called and executed by the processor.

[0086] This embodiment further provides a computer-readable storage medium storing a computer program, and when the computer program is executed by a processor, the above method is implemented.

[0087] Those skilled in the art should understand that the above-mentioned embodiments of the present invention can provide methods, systems or computer program products. Therefore, the present invention can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-readable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. The present invention is described 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 invention. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the functions in the process. Figure 1 A process or multiple processes and / or boxes Figure 1 These computer program instructions can also be stored in a computer-readable memory that can guide a computer or other programmable data processing device to work in a specific way, so that the instructions stored in the computer-readable memory produce a product including an instruction device, which implements the functions specified in the process. Figure 1 A process or multiple processes and / or boxes Figure 1 These computer program instructions can also be loaded onto a computer or other programmable data processing device, so that a series of operation steps are executed on the computer or other programmable device to produce a computer-implemented process, so that the instructions executed on the computer or other programmable device provide for implementing the process in the process. Figure 1 A process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0088] The above is only a preferred embodiment of the present invention, and does not limit the present invention in any form. Although the present invention has been disclosed as a preferred embodiment, it is not intended to limit the present invention. Therefore, any simple modification, equivalent change and modification made to the above embodiment according to the technical essence of the present invention without departing from the content of the technical solution of the present invention shall fall within the scope of protection of the technical solution of the present invention.

Claims

1. An intelligent recommendation method based on hybrid modal opinion consensus decision-making, characterized in that the steps include: Collect the numerical opinions and language opinions of multiple users on the target problem to form a mixed opinion set; The numerical opinion is a numerical value used to represent the degree of the user's emotional tendency towards the target question, and the language opinion is a text used to represent the type of the user's emotional tendency towards the target question; Mixed opinion conversion: converting the numerical opinions in the mixed opinion set into language opinions and converting the language opinions into numerical opinions, respectively obtaining a converted numerical opinion set and a converted language opinion set; when converting the numerical opinions into language opinions, converting the numerical opinions expressed by other users into language opinions to form a unified language opinion format; when converting the language opinions into numerical opinions, converting the language opinions expressed by other users into numerical opinions to form a unified numerical opinion format; Trust weight calculation: Search the user trust set of each user based on the current mixed opinion set, where the user trust set expressing numerical opinions contains numerical opinions whose absolute values ​​of the differences of the numerical opinions of users at different times are within the numerical trust threshold range, and the user trust set expressing language opinions contains language opinions whose absolute values ​​of the differences of the language opinions of users at different times are within the language trust threshold range, and calculate the trust weights between users based on the user trust sets of each user; Mixed opinion evolution: the converted numerical opinion set and language opinion set are evolved. During the evolution process, the numerical opinions and language opinions at the previous moment and the trust weights between the users are used to calculate the numerical opinions and language opinions at the next moment. After the evolution is completed, a new mixed opinion set is formed; The mixed opinion conversion, trust weight calculation and mixed opinion evolution are iteratively performed until the language opinions and numerical opinions in the iterative mixed opinion set reach a consensus, and the recommended result output is determined according to the current consensus state.

2. The intelligent recommendation method based on hybrid modal opinion consensus decision-making according to claim 1 is characterized in that: The numerical opinion Convert to language bigram The expression is: in, It represents the deviation between the user's converted language opinion and numerical opinion. , Express opinions about values The function of Expressing numerical opinion The converted language tuples are Indicates the user's language opinion. represents the approximate rounding function, Indicates the user's language granularity, ; Language bigrams Convert to numerical opinion The expression is: in, Represents the inverse function of the opinion about the value.

3. The intelligent recommendation method based on hybrid modal opinion consensus decision-making according to claim 1 is characterized in that: The searching of the user trust set of each user according to the current mixed opinion set includes: User corresponding to the numerical opinion According to yourself t Numerical opinion and numerical confidence threshold at the moment Find out t User trust set at the moment : in, Indicates to users All users found in t The set of numerical opinions at time, Indicates the found user. Indicates user exist t The numerical opinion at the moment, Indicates user exist t The numerical opinion at the moment, where ; represents the numerical confidence threshold, ; User corresponding to language opinion According to yourself t Language opinion and language trust threshold at the moment Find out t User trust set at the moment : in, Indicates to users All users found in t The collection of language opinions at the moment, Indicates user exist t The language opinions of the moment, Indicates user exist t The language opinions of the moment, , represents the language trust threshold, , Indicates the language granularity of the user.

4. The intelligent recommendation method based on hybrid modal opinion consensus decision-making according to claim 1 is characterized in that: In calculating the trust weights between users according to the user trust sets of each user, the user is calculated according to the following formula: For users Trust weight : in, Indicates user exist t The trust set of other users in their opinions when expressing mixed opinions at any time The number of elements in Indicates user The self-confidence value, Indicates that the user is t A mixed set of opinions at the moment, Trust Set Total number of users in the system.

5. The intelligent recommendation method based on hybrid modal opinion consensus decision-making according to claim 1 is characterized in that: Evolving the converted numerical opinion set and language opinion set includes: The converted numerical opinions are evolved using a numerical opinion evolution model under limited confidence, and the numerical opinion evolution model is: in, , Represents the numerical opinion user exist , t The numerical opinion at the moment, Represents the numerical opinion user exist t The numerical opinion at the moment, Indicates numerical opinion user For users The trust weight of Indicates numerical opinion user The self-confidence value, Indicates numerical opinion user exist t The total number of users in the trust set of other users’ opinions when expressing numerical opinions at a moment; The converted numerical opinions are evolved using a language opinion evolution model under limited confidence, and the language opinion evolution model is: in, , User who expresses language opinion exist , t The language opinions of the moment, Indicates user exist t The language opinions of the moment, User who expresses language opinion For users The trust weight of User who expresses language opinion The self-confidence value, User who expresses language opinion exist t The total number of users in the trust set of other users' opinions when expressing their opinions at a certain moment.

6. The intelligent recommendation method based on hybrid modal opinion consensus decision-making according to any one of claims 1 to 5, characterized in that: The iterative execution of the mixed opinion conversion, trust weight calculation and mixed opinion evolution until the language opinions and numerical opinions in the iterative mixed opinion set reach a consensus includes: Get the mixed opinion set obtained after each iteration; If the obtained mixed opinion set is converted into a numerical opinion set that satisfies the opinion set ,and And converted into a language opinion set to satisfy the opinion set ,and ,in ~ Indicates the various numerical opinions, ~ Express opinions in various languages, It represents the number of numerical opinions or language opinions, and if the deviation of the iteration result value is within the preset allowable range, it is determined that a consensus is reached.

7. An intelligent recommendation device based on hybrid modal opinion consensus decision, characterized in that: include: A mixed opinion collection module is used to collect the numerical opinions and language opinions of multiple users on the target problem to form a mixed opinion set; The numerical opinion is a numerical value used to represent the degree of the user's emotional tendency towards the target question, and the language opinion is a text used to represent the type of the user's emotional tendency towards the target question; A mixed opinion conversion module, used to convert the numerical opinions in the mixed opinion set into language opinions and to convert the language opinions into numerical opinions, respectively obtaining a converted numerical opinion set and a converted language opinion set; when converting the numerical opinions into language opinions, the numerical opinions expressed by other users are converted into language opinions to form a unified language opinion format; when converting the language opinions into numerical opinions, the language opinions expressed by other users are converted into numerical opinions to form a unified numerical opinion format; A trust weight calculation module is used to search for a user trust set of each user based on the current mixed opinion set, wherein the user trust set expressing numerical opinions includes numerical opinions whose absolute values ​​of differences between numerical opinions of users at different times are within a numerical trust threshold range, and the user trust set expressing linguistic opinions includes linguistic opinions whose absolute values ​​of differences between linguistic opinions of users at different times are within a linguistic trust threshold range, and the trust weights between users are calculated based on the user trust sets of each user; A mixed opinion evolution module, used to evolve the converted numerical opinion set and language opinion set, during which the numerical opinions and language opinions at the previous moment and the trust weights between the users are used to calculate the numerical opinions and language opinions at the next moment, and a new mixed opinion set is formed after the evolution is completed; The iterative output module is used to iteratively execute the mixed opinion transformation, trust weight calculation and mixed opinion evolution until the language opinions and numerical opinions in the iterative mixed opinion set reach a consensus, and determine the recommended result output according to the current consensus state.

8. An electronic device comprising a processor and a memory, wherein the memory is used to store a computer program, wherein: The processor is configured to execute the computer program to perform the method according to any one of claims 1 to 6.

9. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 6 is implemented.

Images