Brand value evaluation method and device, equipment and medium

Abstract

The invention provides a brand value evaluation method and device, equipment and a medium, and the method comprises the steps: obtaining text data related to an evaluation brand, and carrying out the preprocessing; classifying and marking the preprocessed text data based on the types of the sound producers, and screening first comment texts of which the sound producers are consumers so as to obtain a first comment text set; performing word segmentation and sentence segmentation on the first comment text to obtain a second comment text set; according to a preset evaluation index system, hit conditions of dimension information related to evaluation indexes in the second evaluation texts are determined and marked, and the dimension information comprises brand information; and screening out target evaluation texts related to the to-be-evaluated brand based on the brand label of the second evaluation text, determining the value of each evaluation index based on the related dimension information in each target evaluation text and the hit condition of each piece of dimension information, and determining the brand value based on the preset weight of each evaluation index.

Description

Technical Field

[0001] The present invention relates to the field of data analysis technology, and in particular to a brand value evaluation method, device, equipment and medium. Background Art

[0002] Amidst increasingly fierce market competition, increasing market transparency, intensified product homogeneity, and growing consumer sophistication, branding has become a key component of a company's intangible assets. With the rapid development of the internet, consumers' active brand commentary on social media has significantly increased, accumulating a vast amount of online data. In the era of big data, this information provides a rich resource for brand valuation.

[0003] To this end, financial and consulting institutions have introduced a variety of analytical tools, including cost methods, financial market methods, and expert evaluation methods, to monitor and quantify brand value. For example, Chinese invention patent CN107093103A discloses a brand value evaluation method based on big data statistical analysis, which includes the use of a method based on big data statistical analysis, taking into account the impact of various factors on brand value, to achieve a rapid and effective evaluation of brand value, and improve user experience. Another Chinese invention patent CN111553609A discloses an Internet brand value evaluation system, which includes experts scoring and summarizing comments for estimation, to achieve an evaluation of the brand's market value and its own strength, and improve the accuracy of the evaluation; however, there are the following problems:

[0004] (1) Lack of consumer psychology perspective: Starting from a market or financial perspective, the measurement of consumer psychology is ignored. Compared with the short-term market perspective, consumer-level psychological indicators can more directly reflect the long-term potential effects of a company's marketing strategy. The existing patent analysis direction cannot accurately assess brand value, because the core of evaluating brand value lies in its value to consumers. In other words, brand value is derived from objective factors such as consumers' cognition, emotional attitudes, and consumption behaviors, and these factors are particularly important for brand customization and differentiated sales.

[0005] (2) Limitations of manual evaluation: Existing expert scoring evaluation methods rely on manual evaluation, which is not only costly but also lacks timeliness and accuracy, making it difficult to provide a quick and effective brand value assessment. In order to comprehensively and accurately measure the brand value perceived by consumers, it is necessary to develop an intelligent evaluation system based on marketing theory.

[0006] In view of this, there is an urgent need to provide a method to evaluate brand value based on the consumer perspective. Summary of the Invention

[0007] In order to overcome the problems existing in the related art, the present disclosure provides a brand value evaluation method, device, equipment and medium to solve the above technical problems in the related art.

[0008] One or more embodiments of this specification provide a method for evaluating brand value, including the following steps:

[0009] Obtaining text data about brand evaluation and preprocessing the text data;

[0010] The preprocessed text data is classified and labeled based on the speaker type, and the first comment texts whose speakers are consumers are selected to obtain the first comment text set;

[0011] Sentence and word segmentation are performed on the first comment text in the first comment text set to obtain a second comment text set;

[0012] According to the preset evaluation index system, determine the hit status of the dimension information related to each evaluation index in each second evaluation text, and mark it as a label on the corresponding second evaluation text, wherein the dimension information includes brand information;

[0013] Based on the brand label of the second evaluation text, target evaluation texts related to the brand to be evaluated are screened out, and the value of each evaluation indicator is determined based on the relevant dimension information in each target evaluation text and the hit situation of each dimension information; then the brand value is determined based on the preset weights of each evaluation indicator.

[0014] Furthermore, after segmenting the first comment text in the first comment text set into words and sentences, the method further includes optimizing each obtained sentence, and the processing steps include:

[0015] Based on the "Ansi Chinese Evaluation Dictionary", confirm whether each short sentence in the sentence contains the text features of the evaluated object. If not, the corresponding short sentence will be merged with the previous short sentence to form a new sentence. Otherwise, it will be treated as a separate sentence until all short sentences are confirmed.

[0016] Furthermore, the dimension information includes brand hits, product attribute hits, fine-grained emotion hits, loyalty behavior hits, intention behavior hits, consumer value orientation hits, cognitive depth classification hits, and purchase stage hits, among which,

[0017] Brand hits are the brand names hit by text features and the number of hits for the corresponding brand names;

[0018] Product attribute hits refer to the product attribute categories hit by text features and the number of hits of the corresponding product attributes;

[0019] The fine-grained sentiment hit situation is the fine-grained sentiment category of the text feature hit and the number of corresponding fine-grained sentiment hits;

[0020] The loyalty behavior hit situation is the loyalty behavior category hit by the text feature and the corresponding loyalty behavior hit count;

[0021] Intention behavior hits, intention behavior categories hit by text features, and the number of corresponding intention behavior hits;

[0022] Consumption value orientation hits, the consumption value orientation categories hit by text features, and the corresponding number of consumption value orientation hits;

[0023] Cognitive depth classification hits, cognitive depth categories of text feature hits and the corresponding number of cognitive depth hits;

[0024] Purchase stage hits, purchase stage categories hit by text features, and the corresponding purchase stage hit count.

[0025] Furthermore, the evaluation indicators and their calculation are specifically as follows: the evaluation indicators include brand prominence, brand recognition depth, product value association, symbolic value association, brand satisfaction and brand loyalty;

[0026] (1) Brand prominence

[0027]

[0028] Among them, k i is the hit status of the brand k to be analyzed on the i-th target evaluation text; WL i is the weighted interaction volume of the i-th target comment text, n is the number of target comment texts, M is the adjustment coefficient, interaction is the interaction volume of a single target comment text, and a is the base of the logarithmic function;

[0029] (2) Depth of brand awareness

[0030]

[0031] Among them, C i is the cognitive depth value corresponding to the cognitive depth category hit by the i-th target comment text, and max() is the behavior of taking the maximum value of all elements in the set in brackets;

[0032] (3) Product value association

[0033]

[0034] Among them, T ijA is the number of text features in the target review text that hit the jth product attribute category under the condition that the intention behavior category of the i-th target review text hits the approximate expression, il The number of text features in the target review text that hit the lth product attribute category under the condition that the intention behavior category of the i-th target review text is avoided;

[0035] (4) Symbolic value association

[0036]

[0037] Among them, V ic R is the number of text features in the target comment text that hit the cth consumption value category under the condition that the hit intention behavior category of the i-th target comment text is approaching, id The number of text features in the target review text that hit the dth consumption value category under the condition that the intention behavior category of the i-th target review text is hit avoided;

[0038] (5) Brand satisfaction

[0039]

[0040] Among them, S ie The number of text features that hit the e-th satisfaction-related emotion for the fine-grained sentiment text features in the i-th target review text, U if The number of text features that match the fth dissatisfaction-related sentiment for the fine-grained sentiment text features in the i-th target review text;

[0041] (6) Brand loyalty

[0042]

[0043] Among them, Loyal i The number of text features that hit the loyalty behavior category for the i-th target comment text, Post i is the number of target review texts whose purchase stage category is after purchase, i Whether the i-th target review text hits the purchase stage category of post-purchase.

[0044] Furthermore, the determination of brand value based on the preset weights of the evaluation indicators specifically includes the following steps:

[0045] The obtained values ​​of each evaluation index are normalized to obtain a standardized value;

[0046] The value of the corresponding brand is determined by taking the preset weights of each indicator and the weighted sum of the standardized values ​​of each evaluation indicator.

[0047] One or more embodiments of this specification provide a method and apparatus for evaluating brand value, including:

[0048] A data acquisition module is used to acquire text data about brand evaluation and pre-process the text data;

[0049] A text screening module is used to classify and label the preprocessed text data based on the speaker type, and screen out the first comment text whose speaker is a consumer, thereby obtaining a first comment text set;

[0050] A word and sentence segmentation processing module is used to perform sentence and word segmentation on the first review text in the first review text set to obtain a second review text set;

[0051] a labeling module, configured to determine, based on a preset evaluation index system, the hit status of dimension information related to each evaluation index in each second evaluation text, and label the corresponding second evaluation text as a label, wherein the dimension information includes brand information;

[0052] The brand value determination module is used to screen out target evaluation texts related to the brand to be evaluated based on the brand label of the second evaluation text, determine the value of each evaluation indicator based on the relevant dimension information in each target evaluation text and the hit situation of each dimension information, and then determine the brand value based on the preset weights of each evaluation indicator.

[0053] Furthermore, it also includes a sentence optimization module;

[0054] The sentence optimization module is used to confirm whether each short sentence in the sentence contains the text features of the evaluated object based on the sentence segmentation results and the "Ansi Chinese Evaluation Dictionary". If not, the corresponding short sentence will be merged with the previous short sentence to form a new sentence. Otherwise, it will be treated as a separate sentence until all short sentences are confirmed.

[0055] Furthermore, the brand value determination module includes a numerical value normalization submodule and a calculation submodule;

[0056] The numerical standardization submodule is used to standardize the obtained values ​​of each evaluation index to obtain a standardized value;

[0057] The calculation submodule determines the value of the corresponding brand based on the preset weights of each indicator and the weighted sum of the standardized values ​​of each evaluation indicator.

[0058] One or more embodiments of this specification provide a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the brand value evaluation method as described in any one of the above items when executing the computer program.

[0059] One or more embodiments of this specification provide a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, the method for evaluating brand value as described in any one of the above items is implemented.

[0060] The present disclosure provides a brand value assessment method, device, equipment, and medium. Advantageously, the method uses marketing concepts such as brand, product, and service, and psychological theories such as consumer emotion, cognition, and behavior to segment collected evaluation text data into words and sentences. The method then determines and annotates each evaluation text and evaluation index based on an evaluation index system that reflects brand value. For a brand to be assessed, the value of each index is determined solely by analyzing target evaluation texts related to the brand, relevant dimensional information within each target evaluation text, and the hits to each dimensional information. Furthermore, the brand value, based on the consumer's mind, is calculated based on preset weights for each evaluation index. The present disclosure utilizes multi-source heterogeneous data to obtain richer brand-related information and mine information from multiple dimensions that can comprehensively assess brand value, providing a more comprehensive and accurate brand value assessment method. The assessment results serve as a basis for exploring the relationships between dimensions, analyzing and determining the brand's current shortcomings and strengths, and providing targeted improvement suggestions for enterprises, thereby enhancing brand value. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] In order to more clearly illustrate one or more embodiments of this specification or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments recorded in this specification. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0062] Figure 1 A flowchart of a brand value evaluation method provided for one or more embodiments of this specification;

[0063] Figure 2 A comparison chart of the scores of various evaluation indicators of a single brand over time under different time nodes / range conditions provided in one or more embodiments of this specification;

[0064] Figure 3 A comparison chart of the scores of various evaluation indicators of multiple brands at the same time point provided in one or more embodiments of this specification;

[0065] Figure 4 A block diagram of a brand value evaluation device provided in one or more embodiments of this specification;

[0066] Figure 5A schematic diagram of the structure of a computer device provided in one or more embodiments of this specification. DETAILED DESCRIPTION

[0067] In order to help those skilled in the art better understand the technical solutions in one or more embodiments of this specification, the technical solutions in one or more embodiments of this specification will be clearly and completely described below in conjunction with the drawings in one or more embodiments of this specification. Obviously, the described embodiments are only part of the embodiments of this specification, not all of the embodiments. Based on one or more embodiments of this specification, all other embodiments obtained by ordinary technicians in this field without making creative efforts should fall within the scope of protection of the present invention.

[0068] The present invention will be described in detail below with reference to specific implementation methods and the accompanying drawings.

[0069] Method Example

[0070] According to an embodiment of the present invention, a method for evaluating brand value is provided, such as Figure 1 FIG. 1 is a flow chart of a method for evaluating brand value provided in this embodiment. The method for evaluating brand value according to an embodiment of the present invention includes the following steps:

[0071] Step S1: Acquire text data about brand evaluation and pre-process the text data.

[0072] In this embodiment, the brand-related consumer-generated content on the Internet constitutes a huge data set, which is the most direct and authentic feedback of consumers on brand awareness. Therefore, the evaluation text data can be collected from social media or OTA platforms, including consumer evaluation and interaction text data on brands. Data preprocessing includes processes such as removing irrelevant noise information and format standardization to ensure data quality.

[0073] Step S2: classify and label the preprocessed text data based on the speaker type, and filter out the first comment texts whose speakers are consumers, thereby obtaining a first comment text set.

[0074] Step S3: Segment the first comment text in the first comment text set into sentences and words to obtain a second comment text set.

[0075] Step S4: According to the preset evaluation index system, the hit status of the dimension information related to each evaluation index in each second evaluation text is determined, and the label is marked on the corresponding second evaluation text, wherein the dimension information includes brand information.

[0076] Step S5: Based on the brand label of the second evaluation text, select target evaluation texts related to the brand to be evaluated, determine the value of each evaluation indicator based on the relevant dimension information in each target evaluation text and the hit status of each dimension information, and determine the brand value based on the preset weights of each evaluation indicator.

[0077] The brand value assessment method provided in this embodiment uses marketing concepts such as brand, product, and service, as well as psychological theories such as consumer emotion, cognition, and behavior, to segment collected evaluation text data into words and sentences. Then, based on an evaluation index system that reflects brand value, each evaluation text and its evaluation index are determined, along with the hits for each dimensional information, and annotated. For the brand to be evaluated, the value of each indicator is determined solely by analyzing the target evaluation text associated with the brand, the relevant dimensional information within each target evaluation text, and the hits for each dimensional information. Based on preset weights for each evaluation index, the brand value based on consumer perception is calculated and determined. This embodiment utilizes multi-source heterogeneous data to obtain richer brand-related information and mine information from multiple dimensions that can comprehensively assess brand value, providing a more comprehensive and accurate brand value assessment method. The assessment results serve as the basis for exploring the relationships between dimensions, analyzing and identifying the brand's current shortcomings and strengths, and providing targeted improvement suggestions for enterprises, thereby enhancing brand value.

[0078] In this embodiment, the multi-source text data collected in step S1 has many problems such as noise or repeated data, so these evaluation text data need to be preprocessed to achieve data validity, accuracy and consistency.

[0079] For example, it includes the integration and normalization of multi-source data, and converts the data into a standard format according to the field standardization rules provided in Table 1. The data contains standardized fields such as "data content", "data type", "release time", "release region", "media source", "number of likes", "number of comments", and "number of reposts". At the same time, each piece of data is numbered as its identifier according to its subordinate relationship, such as "total number", "main post number", etc.

[0080] Table 1. Field standardization rules

[0081] Standardized fields Original fields Data content Comment content | Data content | Content Data Type Data Type | Type Release Time Date|Publish Time|Create Time|Time|pub_time|create_time|cr_time Release Region Source Region | Publishing Region | Region Number of likes Number of likes|number of likes|like|number of likes|like|like Number of comments Number of comments|Number of comments|Comments|Number of replies|Number of replies|Reply|reply Number of forwarding Number of forwardings|Number of forwardings|Forwarding|Number of shares|Shares Number of visits Number of visits|Number of visits|Number of plays|Number of plays|Number of views|Views Number of collections Number of collections|Number of collections|Collections Media Sources Source Website | Media Source Author Name Comment author|Author name|Author|User name|Nickname Author's gender Author gender|gender|sex|user gender

[0082] In this embodiment, the evaluation text data pre-processed in step S1 is classified and labeled based on the speaker type, and the review texts posted by consumers (i.e., the first review texts) are screened out. Because the data itself is multi-source, consumers are separated by account information to filter out relevant review texts from consumers, rather than review texts from self-media, government, etc. This is achieved specifically through the following steps.

[0083] Referring to this step, based on the homepage information of a large number of social media users, as shown in Table 2, the social media user (consumer) categories (as shown in Table 3) are labeled according to the preset label system and an artificial dataset is established. Then, the artificial dataset and natural language text are used to build a social media user classification model, which is finally used to determine the category of the speaker in the comment text big data.

[0084] Table 2. User homepage information classification

[0085]

[0086] Table 3. Social media user categories

[0087]

[0088] The text features of user name A and user-defined description B are transformed into 768-dimensional feature vectors using the sentence transformation model (SBERT-sentence BERT). The first-level and second-level classification data of social media users are transformed into feature vectors of different dimensions using the word transformation model word2vec. The outputs of the sentence transformation model SBERT and the word transformation model word2vec are used as language quantitative expressions, and the social media user type W is identified through a zero-shot learning classifier. * , calculated as follows:

[0089] W * =argmin w {||X T WY|| 2 +λ||WI|| 2}

[0090] Where X is the projection of the SBERT feature vector of the user name, user description, profile content, and user tag text in the word vector space; Y is the word2vec word vector of the first-level classification category and the second-level classification category of the social media user; W is the matrix parameter, λ is a fixed constant, and I is the unit matrix.

[0091] In this embodiment, for the classified data to be analyzed, each user category can be obtained by identification based on the existing classification model. This is an existing technology or a technology well known to those skilled in the art and will not be described in detail here.

[0092] In this embodiment, the pre-processed review text data is segmented into sentences and words to obtain a second evaluation text set to be analyzed. The specific steps for segmenting the first review text are as follows:

[0093] First, the first comment text is segmented by natural language processing. After segmentation encoding, it is stored in the form of a knowledge graph to form a fine-grained knowledge graph. The entries of the fine-grained knowledge graph include feature words, abbreviations, emoticons, symbols, emoticons, sentence patterns and their combinations, which are collectively referred to as text features. The specific implementation includes the following sub-steps:

[0094] In step 311, a semantic dependency algorithm is used to map the first comment text into a graph structure, and a semantic dependency graph is generated for the first comment text to be analyzed. The semantic dependency algorithm can use SDP (Semantic Dependency Parsing) and DEP (Dependency Parsing), which are two important tasks in natural language processing.

[0095] Step 312: Use two graph search algorithms, namely, a depth-first DFS algorithm and a breadth-first BFS algorithm, to search for a suitable language unit LU on the semantic dependency graph. Each language unit LU is a word segmentation result.

[0096] Step 313: Use the N-Gram language model and select the language unit LU={W1, W2, ... W n}; P(W 1:n ) is the joint probability, which is specifically expressed as:

[0097]

[0098] Among them, W l is the text feature unit in the language unit group, l is the order of the text feature unit, n is the upper bound of l, l = 1, 2, ..., n; the relationship between each sentence and its language unit group satisfies the Markov relationship, and the language unit LU = {W1, W2, ...W n}; Each text feature unit W l They are not necessarily connected to each other.

[0099] After word segmentation according to the above steps, the text features to be encoded contained in each first comment text are obtained.

[0100] In this embodiment, step S3 also needs to perform sentence processing on the first comment text in the first comment text set, because after the sentence is divided, a sentence may contain multiple short sentences. For example, "XX company has 6 sub-brands, among which the clothes of XX sub-brand are of excellent quality, and the sports shoes of XX sub-brand are not very popular", such a sentence contains three short sentences; in order to avoid the problem that multiple short sentences point to multiple brand entities, that is, the multi-pointing problem will lead to unclear semantics, or the entity pointed to by the emotional text feature in the sentence is unclear, which leads to inaccurate final analysis and calculation results. Therefore, in order to achieve the unification of brand entities as much as possible, avoid emotions pointing to multiple brands at the same time, so that a sentence (sentence) contains only one brand information as much as possible, it also includes optimizing the obtained sentences. The specific steps are as follows:

[0101] Step 321: Based on the sentence segmentation results and the Ansi Chinese Review Dictionary, determine whether each sentence contains the text features of the evaluated object. If not, the corresponding sentence is merged with the previous sentence to form a new sentence. Otherwise, it is treated as a separate sentence, and the process continues until all sentences are confirmed. The evaluated object can be a brand, a sub-brand, or a product of a brand.

[0102] In this embodiment, word segmentation is stored in the form of a knowledge graph, and then the words contained in a sentence and encoded into certain dimensions can be identified based on the correspondence between the sentences and the words in the sentences, and the first comment texts are also encoded with corresponding dimensions. After the sentence segmentation based on the above method, it is achieved that a sentence contains only one evaluated object as much as possible, avoiding the emotional text features in a sentence from pointing to one brand (evaluated object) as much as possible rather than to multiple brands, and avoiding inaccurate calculation results of subsequent brand value analysis.

[0103] In this embodiment, in step S4, the preset evaluation index system includes evaluation indicators such as brand prominence, brand recognition depth, product value association, symbolic value association, brand satisfaction and brand loyalty, among which brand prominence is characterized by the breadth of the brand's reach and influence on consumers; brand recognition depth is characterized by the accumulated depth of consumers' minds about the brand; product value association is characterized by the relative degree of consumers' discussion of the brand's product content such as product functions, appearance packaging, and price; symbolic value association is characterized by the relative degree of consumers' discussion of the brand's symbolic content such as social value, social value, and self-value; brand satisfaction is characterized by the relative degree to which consumers express various major positive emotions such as love and satisfaction towards the brand; brand loyalty is characterized by the relative degree to which consumers express loyalty intentions and behaviors such as repurchase tendency and recommendation intention towards the brand after purchase.

[0104] In this embodiment, in order to realize the calculation of each evaluation index, it is necessary to determine the relevant dimension information of the evaluation index in each second evaluation text and the hit status of each dimension information based on the knowledge graphs such as "Ansi Chinese Evaluation Dictionary", "Ansi Chinese Emotional Dictionary", and "Ansi Chinese Cognition and Behavior Dictionary" and AI intelligent algorithm models, and mark them as labels on the corresponding second comment text, which may include whether a certain dimension information is hit and the number of hits, and store it as a knowledge graph to obtain the label text. In this embodiment, identifying the hit status of words and sentences in text data based on knowledge graphs and AI intelligent algorithm models is a technology that is easily obtained or known by technical personnel in this field, so it will not be repeated here; among them, the dimension information specifically includes the following.

[0105] A. Hit status and number of hits for each brand

[0106] Based on the sentence and word segmentation results, determine whether each second comment text has a text feature that hits the "brand name" in the "Ansi Chinese Cognition and Behavior Dictionary", the brand names hit by the text feature and the number of hits, and perform corresponding annotation and encoding before storing in the knowledge graph. That is, the annotation and encoding content includes whether the text feature of the comment text hits a certain brand name and the number of corresponding brand names hit. The brand library currently includes but is not limited to more than 2,000 brands. The encoded data theoretically has more than 4,000 brand-related columns.

[0107] B. Hit status and hit count of each product attribute

[0108] Based on the sentence and word segmentation results, determine whether each second comment text has the text features of "product attributes" in the "Ansi Chinese Cognition and Behavior Dictionary", the product attribute categories hit by the text features and the number of hits, and perform corresponding annotations and encoding before storing in the knowledge graph. That is, the annotation and encoding content includes whether the text features of the comment text hit a certain type of product attributes and the number of text features that hit the corresponding category of product attributes. Among them, the product attribute categories in the "Ansi Chinese Cognition and Behavior Dictionary" include: function, price, performance, quality, specifications, packaging, safety risks, technology, raw materials, experience and service attitude.

[0109] C. 50 types of fine-grained emotional hits and hit counts

[0110] Based on the sentence and word segmentation results, determine whether each second comment text has a text feature that matches the "fine-grained emotion" in the "Ansi Chinese Cognition and Behavior Dictionary", the fine-grained emotion category and number of hits, and then perform corresponding annotation and encoding before storing in the knowledge graph. That is, the annotation and encoding content includes whether the text feature of the comment text matches a certain fine-grained emotion and the number of text features that hit the corresponding emotion. The fine-grained emotion categories and 50 emotion words in the "Ansi Chinese Cognition and Behavior Dictionary" are as follows:

[0111] (1) Mood: emotional experiences with low arousal and long duration, including loneliness, numbness, calmness, depression, anxiety, and decadence;

[0112] (2) Stress-related emotions: emotional experiences with high arousal and limited duration, including happiness, fear, surprise, anger, sadness, excitement, relaxation, anxiety, tension, emotion, and alertness;

[0113] (3) Evaluative emotions: The emotional experience generated by an individual after evaluating an information event can be further categorized as directed towards oneself or the outside world. Externally directed emotions include sympathy, indifference, ridicule, disgust, condemnation, resentment, questioning, love, satisfaction, praise, trust, contempt, gratitude, admiration, longing, jealousy, helplessness, expectation, optimism, wishing, and disappointment; internally directed emotions include depression, embarrassment, inferiority, shyness, guilt, pride, frustration, worry, panic, grievance, doubt, and boredom.

[0114] D. Loyalty behavior hit situation and hit number

[0115] Based on the sentence and word segmentation results, we determine whether each second review text matches the "loyalty behavior" category in the "Ansi Chinese Cognition and Behavior Dictionary," as well as the type and number of loyalty behaviors that the text features match. We then annotate and encode the text accordingly and store it in the knowledge graph. The annotation and encoding details include whether the review text features match loyalty behavior and the number of text features that match loyalty behavior. Loyalty behavior words also express loyalty intentions and behaviors after purchase, such as repurchase intention and recommendation willingness.

[0116] E. Hit status and hit count of each intention behavior tendency

[0117] Based on the sentence and word segmentation results, determine whether each second comment text has text features that hit the "intentional behavior" category in the "Ansi Chinese Cognition and Behavior Dictionary", the intentional behavior categories hit by the text features and the number of hits, and then mark and encode them and store them in the knowledge graph. That is, the marking and encoding content includes whether the text features of the comment text hit a certain intentional behavior tendency and the number of text features that hit the corresponding intentional behavior. Among them, the intentional behavior categories in the "Ansi Chinese Cognition and Behavior Dictionary" include: avoidance, approach, wait and see, etc.

[0118] F. Hitting status and number of each consumer value orientation

[0119] Based on the sentence and word segmentation results, determine whether each second comment text has text features that hit the "consumer value orientation" category in the "Ansi Chinese Cognition and Behavior Dictionary", the consumer value orientation category hit by the text features and the number of hits, and perform corresponding annotation and encoding before storing in the knowledge graph. That is, the annotation and encoding content includes whether the text features of the comment text hit a certain consumer value orientation category and the number of text features of the corresponding consumer value orientation category. Among them, the consumer value orientation categories in the "Ansi Chinese Cognition and Behavior Dictionary" include: self-value, social value and social value.

[0120] G. Hit status and hit count of each cognitive depth classification

[0121] Based on the sentence and word segmentation results, determine whether each second comment text has text features that hit the "cognitive depth" category in the "Ansi Chinese Cognition and Behavior Dictionary", the cognitive depth category and the number of hits of the text features, and perform corresponding annotation and encoding before storing in the knowledge graph. That is, the annotation and encoding content includes whether the text features of the comment text hit a certain cognitive depth category and the number of text features that hit the corresponding cognitive depth category. The cognitive depth categories include: perception, understanding, hypothesis reasoning, analogy and comparison, attribution, conclusion and memory.

[0122] H. Hit status and number of hits in each purchase stage

[0123] Based on the sentence and word segmentation results, determine whether each second comment text has text features that hit the "purchase stage category" in the "Ansi Chinese Cognition and Behavior Dictionary", the purchase stage category hit by the text features and the number of hits, and perform corresponding annotation and encoding before storing in the knowledge graph, that is, the annotation and encoding content includes whether the text features of the comment text hit a certain purchase stage and the number of text features that hit a certain purchase stage. Among them, the purchase stage categories in the "Ansi Chinese Cognition and Behavior Dictionary" include: before purchase, during purchase and after purchase.

[0124] In this embodiment, after the word segmentation of each comment text and the hit status of each dimension information of each target comment text are marked, they are stored through the knowledge graph. The advantage is that the knowledge graph is not solidified. As new data is continuously added, the knowledge graph can automatically adjust and optimize itself through the classification model. The data is kept iteratively updated, which can timely reflect consumer attitudes and brand dynamics at the time node and provide the latest brand value evaluation results.

[0125] In this embodiment, based on the hit status and labeling processing of the above-mentioned dimensional information, each comment text with the competition text label is obtained, and then in step S5, based on the brand label, the target evaluation text related to the brand to be evaluated is determined, and then based on the relevant dimensional information in the target text and the hit status of each dimensional information, the relevant value of each evaluation indicator is calculated, among which the evaluation indicators include brand prominence, brand awareness depth, product value association, symbolic value association, brand satisfaction and brand loyalty. The definition and calculation of each evaluation indicator are as follows.

[0126] (1) Brand prominence measures the breadth of a brand's reach and influence on consumers. Based on the target evaluation text, the product of the hit rate of the brand to be analyzed in each target evaluation text and the total number of weighted points of the evaluation text is calculated, and the results of all target evaluation texts are summed to finally obtain the brand prominence of the brand to be analyzed. It reflects the overall satisfaction of consumers with the analyzed brand. The larger the value, the wider the influence of the corresponding brand. The specific calculation formula is as follows:

[0127]

[0128] Among them, k i Is the brand k to be analyzed hit on the target evaluation text of the i-th item, the value is 0 or 1, 0 means no hit, 1 means hit; WL i is the weighted interaction volume of the i-th target comment text, and n is the number of target comment texts.

[0129] The brand library of this embodiment currently includes but is not limited to more than 2,000 brands. The encoded data theoretically contains more than 4,000 brand-related columns.

[0130] In this embodiment, sentiment analysis is performed on the comment text to be analyzed based on the constructed fine-grained sentiment dictionary and sentiment analysis rules, and the number of sentiment text feature hits and the weighted score of the interaction volume of each comment text are determined, thereby determining the sentiment score of each comment text. The specific steps are as follows:

[0131] Use the interaction weighting rule to calculate the weighted interaction volume of a single target comment text. The specific calculation formula is as follows:

[0132]

[0133] Where WL iis the weighted interaction volume of the i-th target comment text. Interaction can be any interactive behavior indicator such as "like", "forward", "comment" in the comment book. M is the adjustment coefficient. When M≤0, the weighted interaction volume formula has no application value. When M>140, the weighted result exceeds the reasonable range and approaches the extreme. Therefore, the value range of M is specified as (0,140]; Interaction is the interaction volume of a single target comment text, Interaction>0; a is the base of the logarithmic function and a>1.

[0134] (2) Brand cognition depth refers to the cumulative depth of consumers’ minds towards the analyzed brand. Based on the target evaluation text, by analyzing the target evaluation text data, the maximum value of consumers’ cognition depth towards the analyzed brand in each target evaluation text is calculated. Then, the maximum value of the cognition depth of all target evaluation texts is averaged to obtain the brand cognition depth value. The larger the value, the deeper the consumer’s cognition towards the analyzed brand. The specific calculation formula is as follows:

[0135]

[0136] Among them, C i A first-order tensor representing the cognitive depth values ​​corresponding to each cognitive depth category hit by the i-th target comment text. Its size represents the cognitive depth value corresponding to the cognitive depth category hit by the i-th target comment text. The specific cognitive depth values ​​corresponding to the cognitive depth categories are shown in Table 4 below; max() represents the behavior of taking the maximum value of all elements in the set in the brackets.

[0137] Table 4. Cognitive behavior categories corresponding to cognitive depth values

[0138]

[0139] (3) Product value association refers to the relative degree of discussion of product attributes such as product function, appearance packaging, and price of the analyzed brand by consumers. Based on the target evaluation text, by analyzing the target evaluation text data, the product attribute hit of the analyzed brand in each target evaluation text is calculated (the value is 0 (missing) or 1 (hit)), the total weighted points of the target evaluation text, and the sum of the hits of each intention behavior word are multiplied, and the results of all target evaluation texts are summed. Finally, the ratio of the sum of the hits of any intention behavior table "approaching" words to the sum of the hits of all intention behavior words is calculated. It reflects the relative degree of discussion of product content such as product function, appearance packaging, and price of the analyzed brand by consumers. The larger the value, the more consumers tend to "approach" the product content of the analyzed brand. The specific calculation is as follows:

[0140]

[0141] Among them, T and A are both second-order tensors with sizes of n×11, T ij A is the number of text features in the target review text that hit the jth product attribute category under the condition that the intention behavior category of the i-th target review text hits the approximate expression, il k is the number of text features in the target review text that hit the lth product attribute category under the condition that the intention behavior category hits avoidance. The words for intention behavior include: avoidance, approach, and wait and see. The product attributes include related words such as function, price, performance, quality, specifications, packaging, safety risk, technology, raw materials, experience, and service attitude. i is the hit status of the brand k to be analyzed on the i-th target review text, and the value is 0 or 1.

[0142] (4) Symbolic value association refers to the relative degree of discussion of the social value, social value, self-value and other symbolic contents of the analyzed brand by consumers based on the target evaluation text. The specific calculation is as follows:

[0143]

[0144] Among them, V and R are both second-order tensors, both of size n×3, V ic R is the number of text features in the target comment text that hit the cth consumption value category under the condition that the hit intention behavior category of the i-th target comment text is approaching, id The number of text features in the target review text that hit the dth consumption value category under the condition that the intention behavior category of the i-th target review text is avoided.

[0145] (5) Brand satisfaction refers to the overall satisfaction of consumers with the analyzed brand, which is reflected by their fine-grained emotional responses such as liking and satisfaction. This is obtained by analyzing multiple evaluation text data, calculating the hit count of the analyzed brand in each evaluation text, the total number of weighted points of the evaluation text, and the product of the sum of the hit counts of satisfaction-related emotions and dissatisfaction-related emotions, and summing the results of all evaluation texts. It reflects the overall satisfaction of consumers with the analyzed brand. The larger the value, the higher the consumer's satisfaction with the brand. The specific calculation formula is as follows:

[0146]

[0147] Among them, S ie and U if Both are second-order tensors of size n×5, S ie The number of text features that hit the e-th satisfaction-related emotion for the fine-grained sentiment text features in the i-th target review text, U ifThe number of text features of the fth dissatisfaction-related emotion that are hit by the fine-grained sentiment text features in the i-th target review text, where satisfaction-related emotions are the set Satis = {like, satisfaction, praise, happiness, excitement}, and dissatisfaction-related emotions are the set Unsatis = {disgust, anger, condemnation, sadness, disappointment}.

[0148] (6) Brand loyalty refers to the relative degree to which consumers express loyalty intentions and behaviors such as repurchase tendency and recommendation willingness after purchasing the brand to be analyzed. By analyzing the target review text data, the hit situation of the brand to be analyzed k in each target evaluation text, the total weighted points of the target evaluation text, and the product of the sum of the hit situation of loyalty behavior and post-purchase expression are calculated, and the results of all target review texts are summed. Finally, the ratio of the sum of the hit number of loyalty behavior to the sum of the hit number of loyalty behavior and post-purchase expression is calculated. The larger the value, the higher the consumer's loyalty to the brand. The specific calculation is as follows:

[0149]

[0150] Among them, Loyal i The number of text features representing the loyalty behavior category of the i-th target comment text, Post i is the number of target review texts whose purchase stage category is after purchase. i It is whether the target review text of the i-th item hits the purchase stage category of post-purchase. The value is 0 or 1, 0 means miss and 1 means hit, which means that the target review text is a review expressed after purchase.

[0151] In this embodiment, the brand value is determined based on the calculation results of the above-mentioned evaluation indicators and the preset weights of the evaluation indicators, which is specifically achieved through the following steps:

[0152] Step 51, normalizing the obtained values ​​of each evaluation index to obtain a normalized value;

[0153] Since the calculation results of the above-mentioned evaluation indicators are different values, for example, the calculation results of brand satisfaction and prominence may be tens of values, while indicators such as brand awareness depth, product value association, and loyalty may be decimal values. Therefore, it is necessary to standardize each evaluation indicator to obtain standardized data, which also facilitates the visualization of the indicators.

[0154] Step 52: Determine the value of the corresponding brand by taking the weighted sum of the preset weights of each indicator and the standardized values ​​of each evaluation indicator. The specific calculation is as follows:

[0155]

[0156] Among them, T k is the total score of brand k; B is a second-order tensor representing the weight of the evaluation index h of brand k; Z is a second-order tensor representing the value of the evaluation index h of brand k.

[0157] In a specific embodiment, the weight values ​​of the evaluation indicators brand prominence, brand recognition depth, product value association, symbolic value association, brand satisfaction and brand loyalty are set to 0.12, 0.18, 0.12, 0.18, 0.16 and 0.24 respectively.

[0158] The method of this embodiment is based on the theory of brand equity and constructs a set of quantitative index systems, including prominence, cognitive depth, product value association, symbolic value association, satisfaction, loyalty, etc., to comprehensively evaluate the brand value from the perspective of consumers. Based on this evaluation system, the most direct and real information on brand cognition feedback from consumers on the Internet is analyzed, and combined with natural language processing and machine learning technology, the relationship between the brand and related entities is accurately identified and quantified, thereby improving the accuracy of the evaluation, and referring to Figure 2 and Figure 3 As shown, it is to use different time nodes / ranges as grouping fields, calculate the six-dimensional scores of each group, and thus compare the six evaluation index scores of a single brand at different time nodes / ranges, and use different brands as grouping fields, calculate the six evaluation index scores of each group at the same time node, and thus determine the brand value comparison of different brands. This solves the problem that existing tools rely on manual scoring, which is not only time-consuming and labor-intensive, but also cannot be used to compare the values ​​of multiple brands due to its strong subjectivity.

[0159] Device embodiment

[0160] According to an embodiment of the present invention, a device for evaluating brand value is provided. Figure 4 FIG. 1 is a block diagram of a brand value evaluation device according to an embodiment of the present invention. The brand value evaluation device according to an embodiment of the present invention includes:

[0161] The data acquisition module 10 is used to acquire text data related to brand evaluation and pre-process the text data.

[0162] The text screening module 20 is used to classify and label the pre-processed text data based on the speaker type, and screen out the first comment texts whose speakers are consumers, thereby obtaining a first comment text set.

[0163] The word and sentence segmentation processing module 30 is used to perform sentence and word segmentation on the first review text in the first review text set to obtain a second review text set.

[0164] The marking module 40 is used to determine the hit status of the dimension information related to each evaluation indicator in each second evaluation text according to the preset evaluation indicator system, and mark it as a label on the corresponding second evaluation text, where the dimension information includes brand information.

[0165] The brand value determination module 50 is used to screen out target evaluation texts related to the brand to be evaluated based on the brand label of the second evaluation text, determine the value of each evaluation indicator based on the relevant dimension information in each target evaluation text and the hit situation of each dimension information, and determine the brand value based on the preset weights of each evaluation indicator.

[0166] The brand value assessment device provided in this embodiment uses marketing concepts such as brand, product, and service, and psychological theories such as consumer emotion, cognition, and behavior to segment collected evaluation text data into words and sentences. Then, based on an evaluation index system that can reflect brand value, the word segmentation and sentence processing module 30 and the annotation module 40 respectively determine the evaluation index and the hit status of each dimensional information for each evaluation text, and annotate them. For the brand to be evaluated, it is only necessary to determine the value of each indicator based on analyzing the target evaluation text related to the brand and the relevant dimensional information in each target evaluation text and the hit status of each dimensional information. The brand value is determined based on the preset weights of each evaluation index. The method of this embodiment utilizes multi-source heterogeneous data to obtain richer brand-related information and mine information from multiple dimensions that can comprehensively evaluate brand value, providing a more comprehensive and accurate brand value assessment method. The assessment results serve as the basis for exploring the relationship between each dimension, analyzing and determining the current shortcomings and advantages of the brand, and providing targeted improvement suggestions for the enterprise, thereby improving brand value.

[0167] In this embodiment, a sentence optimization module is also provided to confirm whether each short sentence in the sentence contains the text features of the evaluated object based on the sentence segmentation results and the "Ansi Chinese Evaluation Dictionary". If not, the corresponding short sentence is merged with its previous short sentence to form a new sentence; otherwise, it is treated as a separate sentence until all short sentences are confirmed.

[0168] In this embodiment, after the word segmentation of each comment text and the hit status of each dimension information of each target comment text are annotated, they are stored through the knowledge graph. In the existing technology, knowledge graphing can be achieved through entity / relationship extraction → triple mapping → graph database storage of labeled text data.

[0169] In this embodiment, the dimension information specifically includes the following:

[0170] The hit status and hit count of each brand are: whether the text of each second comment contains the text feature of "brand name" in the "Ansi Chinese Cognition and Behavior Dictionary", the brand name hit by the text feature, and the number of hits;

[0171] The hit status and hit count of each product attribute, including whether the text of each second comment contains a text feature that matches the term "product attribute" in the "Ansi Chinese Cognition and Behavior Dictionary", the product attribute category hit by the text feature, and the number of hits;

[0172] The fine-grained emotion hit status and hit count are: whether each second comment text has a hit on the "fine-grained emotion" text feature in the "Ansi Chinese Cognition and Behavior Dictionary", the fine-grained emotion category hit by the text feature, and the hit count;

[0173] The hit status and hit count of loyalty behavior are: whether each second comment text has a hit on the text feature of the "loyalty behavior" category in the "Ansi Chinese Cognition and Behavior Dictionary", the loyalty behavior category hit by the text feature, and the hit count;

[0174] The hit status and hit count of each intentional behavior tendency are as follows: whether each second comment text has a hit on the "intentional behavior" category text feature in the "Ansi Chinese Cognition and Behavior Dictionary", the intentional behavior category hit by the text feature, and the hit count;

[0175] The hit status and hit count of each consumer value orientation are: whether each second comment text has a hit text feature of the "consumer value orientation" category in the "Ansi Chinese Cognition and Behavior Dictionary", the consumer value orientation category hit by the text feature, and the hit count;

[0176] The hit status and hit count of each cognitive depth classification is to determine whether each second comment text has a hit on the "cognitive depth" category text feature in the "Ansi Chinese Cognition and Behavior Dictionary", the cognitive depth category hit by the text feature, and the hit count;

[0177] The hit status and hit count for each purchase stage are whether each second comment text has a hit on the text feature of the "Purchase Stage Category" in the "Ansi Chinese Cognition and Behavior Dictionary", the purchase stage category hit by the text feature, and the hit count.

[0178] In this embodiment, the brand value determination module 50 determines the target evaluation text related to the brand to be evaluated based on the brand tag, and then calculates the relevant value of each evaluation indicator based on the relevant dimension information and the hit status of each dimension information in the target text. The definition and calculation of each evaluation indicator are as follows:

[0179] (1) Brand salience measures the breadth of a brand’s reach and influence on consumers. By analyzing the target evaluation texts, the product of the hit count of the brand to be analyzed in each target evaluation text, the total number of weighted points of the evaluation text, and the sum of the hit counts of satisfaction-related and dissatisfaction-related emotions is calculated. The results of all target evaluation texts are summed to finally obtain the satisfaction of the brand to be analyzed. The specific calculation is as follows:

[0180]

[0181] Among them, k i Is the brand k to be analyzed hit on the target evaluation text of the i-th item, the value is 0 or 1, 0 means no hit, 1 means hit; WL i is the weighted interaction volume of the i-th target comment text, and n is the number of target comment texts.

[0182] In this embodiment, sentiment analysis is performed on the comment text to be analyzed based on the constructed fine-grained sentiment dictionary and sentiment analysis rules, and the number of sentiment text feature hits and the weighted score of the interaction volume of each comment text are determined, thereby determining the sentiment score of each comment text. The specific steps are as follows:

[0183] Use the interaction weighting rule to calculate the weighted interaction volume of a single target comment text. The specific calculation formula is as follows:

[0184]

[0185] Where WL i The sentiment k of the i-th target comment text j The weighted interaction volume of a comment can be any interactive behavior indicator such as "like," "repost," or "comment." M is the adjustment coefficient for sentiment. When M ≤ 0, the formula has no practical value. When M > 140, the weighted result exceeds the reasonable range and approaches an extreme. Therefore, the value range of M is specified as (0, 140). Interaction is the interaction volume of a single target comment text, and Interaction > 0. a is the base of the logarithmic function and a > 1.

[0186] (2) Brand cognition depth refers to the cumulative depth of consumers’ minds regarding the analyzed brand. By analyzing the target evaluation text data, the maximum value of consumers’ cognition depth regarding the analyzed brand in each target evaluation text is calculated. Then, the maximum values ​​of the cognition depth of all target evaluation texts are averaged to obtain the brand cognition depth value. The specific calculation formula is as follows:

[0187]

[0188] Among them, C i A first-order tensor representing the cognitive depth values ​​corresponding to each cognitive depth category hit by the i-th target comment text. Its size represents the cognitive depth value corresponding to the cognitive depth category hit by the i-th target comment text. max() represents the behavior of taking the maximum value of all elements in the set in the brackets.

[0189] (3) Product value association refers to the relative degree of discussion among consumers on the product attributes of the analyzed brand, such as product function, appearance and packaging, and price. By analyzing multiple target evaluation text data, the product attribute hits of the analyzed brand in each target evaluation text (the value is 0 or 1), the total number of weighted points of the target evaluation text, and the sum of the hits of each intention behavior word are calculated. The results of all target evaluation texts are summed up, and finally the ratio of the sum of the hits of any intention behavior table "approaching" words to the sum of the hits of all intention behavior words is calculated. The specific calculation is as follows:

[0190]

[0191] Among them, T and A are both second-order tensors with sizes of n×11, T ij A is the number of text features in the target review text that hit the jth product attribute category under the condition that the intention behavior category of the i-th target review text hits the approximate expression, il k is the number of text features in the target review text that hit the lth product attribute category under the condition that the intention behavior category hits avoidance. The words for intention behavior include: avoidance, approach, and wait and see. The product attributes include related words such as function, price, performance, quality, specifications, packaging, safety risk, technology, raw materials, experience, and service attitude. i is the hit status of the brand k to be analyzed on the i-th target review text, and the value is 0 or 1.

[0192] (4) Symbolic value association refers to the relative degree of discussion of the brand's social value, social value, self-value and other symbolic contents by consumers. The specific calculation is as follows:

[0193]

[0194] Among them, V and R are both second-order tensors with a size of n×3, V ic R is the number of text features in the target comment text that hit the cth consumption value category under the condition that the hit intention behavior category of the i-th target comment text is approaching, id The number of text features in the target review text that hit the dth consumption value category under the condition that the intention behavior category of the i-th target review text is avoided.

[0195] (5) Brand satisfaction refers to the overall satisfaction of consumers with the analyzed brand, which is reflected in their fine-grained emotional responses such as liking and satisfaction. This is achieved by analyzing multiple evaluation texts, calculating the hit count of the analyzed brand in each evaluation text, the total number of weighted points of the evaluation text, and the product of the sum of the hit counts of satisfaction-related emotions and dissatisfaction-related emotions, and summing the results of all evaluation texts. The specific calculation is as follows:

[0196]

[0197] Among them, S ij and U ij are all second-order tensors of size n×5, S ie The number of text features that hit the e-th satisfaction-related emotion for the fine-grained sentiment text features in the i-th target review text, U if The number of text features of the fth dissatisfaction-related emotion that are hit by the fine-grained sentiment text features in the i-th target review text, where satisfaction-related emotions are the set Satis = {like, satisfaction, praise, happiness, excitement}, and dissatisfaction-related emotions are the set Unsatis = {disgust, anger, condemnation, sadness, disappointment}.

[0198] (6) Brand loyalty refers to the relative degree to which consumers express loyalty intentions and behaviors such as repurchase tendency and recommendation intention after purchasing the brand to be analyzed. By analyzing multiple target review text data, the hit situation of the brand to be analyzed k in each target evaluation text, the total weighted points of the target evaluation text, and the product of the sum of the hit situation of loyalty behavior and post-purchase expression are calculated, and the results of all target review texts are summed. Finally, the ratio of the sum of the hit number of loyalty behavior to the sum of the hit number of loyalty behavior and post-purchase expression is calculated. The specific calculation is as follows:

[0199]

[0200] Among them, Loyal i Represents the number of text features that hit the loyalty behavior category for the i-th target comment text, Post i is the number of target review texts whose purchase stage category is after purchase. i It is whether the i-th target review text hits the purchase stage category of post-purchase.

[0201] In this embodiment, since the calculation results of the above-mentioned evaluation indicators are different values, for example, the calculation results of brand satisfaction and prominence may be tens of values, while the indicators such as brand recognition depth, product value association, and loyalty may be decimal values, it is necessary to standardize the evaluation indicators to obtain standardized data, which also facilitates the visualization of the indicators. Therefore, the brand value determination module 50 includes a value standardization submodule and a calculation submodule.

[0202] The numerical standardization submodule is used to standardize the obtained values ​​of each evaluation index to obtain a standardized value;

[0203] The calculation submodule determines the value of the corresponding brand based on the preset weights of each indicator and the weighted sum of the standardized values ​​of each evaluation indicator. The specific calculation is as follows:

[0204]

[0205] Among them, T k is the total score of brand k; B is a second-order tensor representing the weight of the evaluation index h of brand k; Z is a second-order tensor representing the value of the evaluation index h of brand k.

[0206] The embodiment of the present invention is an apparatus embodiment corresponding to the above-mentioned method embodiment. The specific operations of the processing steps of each module can be understood by referring to the description of the method embodiment, and will not be repeated here.

[0207] like Figure 5 As shown, the present invention also provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the brand value evaluation method in the above embodiment is implemented.

[0208] The present invention also provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the evaluation of the brand value in the above embodiment is implemented, or when the computer program is executed by a processor, the evaluation method of the brand value in the above embodiment is implemented.

[0209] Those skilled in the art will appreciate that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0210] Each embodiment in this specification is described in a progressive manner. The same or similar parts between the embodiments can be referred to each other. Each embodiment focuses on the differences from other embodiments. In particular, for the device or system embodiments, since they are basically similar to the method embodiments, the description is relatively simple. For the relevant parts, refer to the partial description of the method embodiments. The device and system embodiments described above are merely schematic, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the scheme of this embodiment. A person of ordinary skill in the art can understand and implement it without making any creative efforts.

[0211] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the scope of the technical solutions of the embodiments of the present invention, and the contents not described in detail in the specification of the present invention are common knowledge to those skilled in the art.

Claims

1. A method for evaluating brand value, characterized in that: Including steps: Obtaining text data about brand evaluation and preprocessing the text data; The preprocessed text data is classified and labeled based on the speaker type, and the first comment texts whose speakers are consumers are selected to obtain the first comment text set; Sentence and word segmentation are performed on the first comment text in the first comment text set to obtain a second comment text set; According to the preset evaluation index system, determine the hit status of the dimension information related to each evaluation index in each second evaluation text, and mark it as a label on the corresponding second evaluation text, wherein the dimension information includes brand information; Based on the brand label of the second evaluation text, target evaluation texts related to the brand to be evaluated are screened out, and the value of each evaluation indicator is determined based on the relevant dimension information in each target evaluation text and the hit situation of each dimension information; then the brand value is determined based on the preset weights of each evaluation indicator.

2. The brand value evaluation method according to claim 1, wherein: After the first comment text in the first comment text set is segmented into sentences and words, the method further includes optimizing each obtained sentence, and the processing steps include: Based on the knowledge graph and AI intelligent algorithm, confirm whether each short sentence in the sentence contains the text features of the evaluated object. If not, the corresponding short sentence will be merged with its previous short sentence to form a new sentence. Otherwise, it will be treated as a separate sentence until all short sentences are confirmed.

3. The brand value evaluation method according to claim 1, wherein: The dimension information includes brand hits, product attribute hits, fine-grained emotion hits, loyalty behavior hits, intention behavior hits, consumer value orientation hits, cognitive depth classification hits, and purchase stage hits, among which, Brand hits are the brand names hit by text features and the number of hits for the corresponding brand names; Product attribute hits refer to the product attribute categories hit by text features and the number of hits of the corresponding product attributes; The fine-grained sentiment hit situation is the fine-grained sentiment category of the text feature hit and the number of corresponding fine-grained sentiment hits; The loyalty behavior hit situation is the loyalty behavior category hit by the text feature and the corresponding loyalty behavior hit count; Intention behavior hits, intention behavior categories hit by text features, and the number of corresponding intention behavior hits; Consumption value orientation hits, the consumption value orientation categories hit by text features, and the corresponding number of consumption value orientation hits; Cognitive depth classification hits, cognitive depth categories of text feature hits and the corresponding number of cognitive depth hits; Purchase stage hits, purchase stage categories hit by text features, and the corresponding purchase stage hit count.

4. The brand value evaluation method according to claim 3, wherein: The evaluation indicators and their calculation are as follows: the evaluation indicators include brand prominence, brand recognition depth, product value association, symbolic value association, brand satisfaction and brand loyalty; (1) Brand prominence Among them, k i is the hit status of the brand k to be analyzed on the i-th target evaluation text; WL i is the weighted interaction volume of the i-th target comment text, n is the number of target comment texts, M is the adjustment coefficient, interaction is the interaction volume of a single target comment text, and a is the base of the logarithmic function; (2) Depth of brand awareness Among them, C i is the cognitive depth value corresponding to the cognitive depth category hit by the i-th target comment text, and max() is the behavior of taking the maximum value of all elements in the set in brackets; (3) Product value association Among them, T ij A is the number of text features in the target review text that hit the jth product attribute category under the condition that the intention behavior category of the i-th target review text hits the approximate expression, il The number of text features in the target review text that hit the lth product attribute category under the condition that the intention behavior category of the i-th target review text is avoided; (4) Symbolic value association Among them, V ic R is the number of text features in the target comment text that hit the cth consumption value category under the condition that the hit intention behavior category of the i-th target comment text is approaching, id The number of text features in the target review text that hit the dth consumption value category under the condition that the intention behavior category of the i-th target review text is hit avoided; (5) Brand satisfaction Among them, S ie The number of text features that hit the e-th satisfaction-related emotion for the fine-grained sentiment text features in the i-th target review text, U if The number of text features that hit the fth dissatisfaction-related sentiment for the fine-grained sentiment text features in the i-th target review text; (6) Brand loyalty Among them, Loyal i The number of text features that hit the loyalty behavior category for the i-th target comment text, Post i is the number of target review texts whose purchase stage category is after purchase, i Whether the i-th target review text hits the purchase stage category of post-purchase.

5. The brand value evaluation method according to any one of claims 1 to 4, characterized in that: The determination of brand value based on the preset weights of the evaluation indicators specifically includes the following steps: The obtained values ​​of each evaluation index are normalized to obtain a standardized value; The value of the corresponding brand is determined by taking the preset weights of each indicator and the weighted sum of the standardized values ​​of each evaluation indicator.

6. A brand value evaluation device, characterized in that: include: A data acquisition module is used to acquire text data about brand evaluation and pre-process the text data; A text screening module is used to classify and label the preprocessed text data based on the speaker type, and screen out the first comment text whose speaker is a consumer, thereby obtaining a first comment text set; A word and sentence segmentation processing module is used to perform sentence and word segmentation on the first review text in the first review text set to obtain a second review text set; a labeling module, configured to determine, based on a preset evaluation index system, the hit status of dimension information related to each evaluation index in each second evaluation text, and label the corresponding second evaluation text as a label, wherein the dimension information includes brand information; The brand value determination module is used to screen out target evaluation texts related to the brand to be evaluated based on the brand label of the second evaluation text, determine the value of each evaluation indicator based on the relevant dimension information in each target evaluation text and the hit situation of each dimension information, and then determine the brand value based on the preset weight calculation of each evaluation indicator.

7. The brand value evaluation device according to claim 6, wherein: It also includes a sentence optimization module; The sentence optimization module is used to confirm whether each short sentence in the sentence contains the text features of the evaluated object based on the sentence segmentation results, knowledge graph and AI intelligent algorithm. If not, the corresponding short sentence will be merged with its previous short sentence to form a new sentence. Otherwise, it will be treated as a separate sentence until all short sentences are confirmed.

8. The brand value evaluation device according to claim 6, wherein: The brand value determination module includes a numerical value normalization submodule and a calculation submodule; The numerical standardization submodule is used to standardize the obtained values ​​of each evaluation index to obtain a standardized value; The calculation submodule determines the value of the corresponding brand based on the preset weights of each indicator and the weighted sum of the standardized values ​​of each evaluation indicator.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the brand value evaluation method according to any one of claims 1 to 5 is implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the brand value evaluation method according to any one of claims 1 to 5 is implemented.

Images