A method, apparatus and storage medium for evaluating an advertisement

Abstract

This invention provides a method, apparatus, and storage medium for evaluating advertisements. The method includes: preprocessing and structuring multi-source heterogeneous brand data to generate triple data; performing entity extraction, relation extraction, and multi-dimensional knowledge fusion on the triple data to construct a brand-specific knowledge graph; labeling successfully matched nodes and semantic relationship edges with corresponding signal levels and attaching verification metadata according to a preset E-E-A-T authoritative signal rule library; extracting experience signal values, professionalism signal values, authority signal values, and credibility signal values ​​from the brand-specific knowledge graph based on the signal levels and the verification metadata; and calculating an overall authority index using a multi-factor dynamic weighting algorithm on the experience signal values, professionalism signal values, authority signal values, and credibility signal values ​​to evaluate the target advertisement. This method improves the accuracy of advertisement evaluation.

Description

Technical Field

[0001] This invention relates to the fields of digital marketing, artificial intelligence and information retrieval technology, and specifically to a method, apparatus and storage medium for evaluating advertisements. Background Technology

[0002] With the rapid popularization of generative artificial intelligence (AI) technology, AI has become a new core entry point for users to obtain information, and traditional search engine optimization (SEO) is gradually evolving towards generative engine optimization (GEO). Currently, the industry generally uses the EEAT principle as the main evaluation standard for the quality of generative AI content and the credibility of sources, focusing on the experience, professionalism, authority, and credibility of the content, providing a basis for AI to identify information and output answers.

[0003] Currently, two prominent industry problems exist in generative engine applications, severely disrupting the quality of AI information output and market dissemination order. First, malicious GEO optimization is rampant. Some businesses obfuscate the source of information for generative engines by mass-producing advertorials, false evaluations, and exaggerated claims, interfering with AI's source identification logic. Inferior and false sources easily mislead AI judgment, infringing not only on users' right to know but also undermining fair industry competition. Second, the utilization rate of corporate brand authority assets is low. Core brand assets such as awards, certifications, patents, and case studies are often stored in unstructured or semi-structured formats like text documents and web pages, lacking a unified and standardized format. Generative AI struggles to accurately capture and interpret authoritative signals, resulting in issues such as missing information, content errors, and weakened authority in AI-generated answers.

[0004] To address the aforementioned issues, existing technical solutions have significant shortcomings. Current mainstream technologies can only achieve single-function industry knowledge graph construction or information trust assessment, resulting in limited functionality. Existing solutions cannot complete an integrated process of structuring and organizing brand assets, standardizing authoritative signals, and optimizing for AI adaptation; they also lack the technical capability to proactively push brand authority attributes to generative artificial intelligence.

[0005] In summary, existing technologies cannot solve industry challenges such as malicious GEO interference, difficulties in extracting brand authority signals, and distortion of AI brand representation, making them ill-suited for the brand optimization and information source purification needs of the generative engine era. Therefore, there is an urgent need to develop an integrated generative engine brand authority optimization solution to fill the current technological gap. Summary of the Invention

[0006] The purpose of this invention is to provide a method, apparatus, and storage medium for evaluating advertisements, which improves the accuracy of advertisement evaluation.

[0007] To achieve the above objectives, embodiments of the present invention provide a method for evaluating advertisements. The method includes: collecting multi-source heterogeneous brand data associated with the target advertisement; preprocessing and structuring the multi-source heterogeneous brand data to generate triplet data; performing entity extraction, relation extraction, and multi-dimensional knowledge fusion on the triplet data based on an ontology customized for the brand marketing scenario to construct a brand-specific knowledge graph; traversing all entity nodes and semantic relationship edges of the brand-specific knowledge graph, and according to a preset EEAT authoritative signal rule base, labeling successfully matched nodes and semantic relationship edges with corresponding signal levels and attaching verification metadata; extracting experience signal values, professionalism signal values, authority signal values, and credibility signal values ​​from the brand-specific knowledge graph based on the signal levels and the verification metadata; and calculating an overall authority index using a multi-factor dynamic weighting algorithm on the experience signal values, professionalism signal values, authority signal values, and credibility signal values ​​to evaluate the target advertisement.

[0008] Optionally, the subject of the triple data includes brand, founder, product, and industry organization; the relationship of the triple data includes awards received, patents owned, being reported by authoritative media, and professional qualifications; the attributes of the triple data include years of experience, certification level, number of cases, and official endorsement sources.

[0009] Optionally, the step of traversing all entity nodes and semantic relationship edges of the brand-specific knowledge graph, and labeling the successfully matched nodes and semantic relationship edges with corresponding signal levels and attaching verification metadata according to the preset EEAT authoritative signal rule library, includes: traversing all entity nodes and semantic relationship edges of the brand-specific knowledge graph using a depth-first search algorithm, extracting node attributes of each node, and extracting the relationship type of each edge; setting the EEAT authoritative signal rule library, including trigger conditions, signal types, signal levels, and verification metadata; matching the node attributes and relationship types with the trigger conditions in the EEAT authoritative signal rule library; and labeling and attaching verification metadata to the successfully matched nodes and semantic relationship edges according to the signal type and signal level, including: basic evidence information, verification status, and evidence chain ID.

[0010] Optionally, the overall authority index is: A=(α×E+β×Ex+γ×Au+δ×Tr)×CP Where A is the overall authority index, E is the empirical signal value, Ex is the professional signal value, Au is the authority signal value, Tr is the credibility signal value, C is the signal level correction coefficient, P is the negative information penalty factor, and α, β, γ, and δ are the weight coefficients of empirical signal, professional signal, authority signal, and credibility signal, respectively, and α+β+γ+δ=1.

[0011] Optionally, the following parameters are defined: Experience Signal Value = Original Experience Signal Value / Highest Industry Experience Signal Value; Original Experience Signal Value = All Customer Case Values ​​+ All User Empirical Values ​​+ Application Timeline Values ​​+ All Quantitative Effect Indicator Values; Customer Case Value = Customer Level Coefficient × Cooperation Length Coefficient × Effect Quantification Coefficient - Customer Level Coefficient; Professionalism Signal Value = Original Professionalism Signal Value / Highest Industry Professionalism Signal Value; Original Professionalism Signal Value = All Qualification Certification Values ​​+ All Patent Values ​​+ All Industry Standard Values; Qualification Certification Value = Certification Level Coefficient × Validity Coefficient × Official Verification Coefficient - Certification Level Coefficient; Authority Signal Value = Original Authority Signal Value / Highest Industry Authority Signal Value; Original Authority Signal Value = All Authoritative Media Reports Values ​​+ All Third-Party Agency Rating Values ​​+ Cross-Platform Consistency Value; Credibility Signal Value = max(0, Original Credibility Signal Value / 10); Original Credibility Signal Value = Information Timeliness Value + Information Transparency Value + Information Consistency Value - Negative Information Value.

[0012] Optionally, the experience signals include linked customer cases, user empirical data, application timelines, and quantitative performance indicators; the professional signals include associated qualification certifications, patents, industry standards, and official verification links; the authoritative signals include marking the level of third-party authoritative sources, reporting time, reprint scope, and cross-platform consistency; and the credibility signals include integrating information timeliness, transparency, and consistency to generate basic authoritative labels.

[0013] Optionally, the method further includes: the preprocessing includes cleaning, deduplication, and formatting of multi-source heterogeneous brand data.

[0014] On the other hand, this application also proposes an apparatus for evaluating advertisements, comprising: an acquisition module for collecting multi-source heterogeneous brand data associated with the target advertisement, and preprocessing and structuring the multi-source heterogeneous brand data to generate triple data; a first processing module for performing entity extraction, relation extraction, and multi-dimensional knowledge fusion on the triple data based on an ontology customized for the brand marketing scenario, for constructing a brand-specific knowledge graph; a second processing module for traversing all entity nodes and semantic relation edges of the brand-specific knowledge graph, and labeling the successfully matched nodes and semantic relation edges with corresponding signal levels and attaching verification metadata according to a preset EEAT authoritative signal rule base; a third processing module for extracting experience signal values, professionalism signal values, authority signal values, and credibility signal values ​​from the brand-specific knowledge graph according to the signal levels and the verification metadata; and a fourth processing module for calculating an overall authority index by performing a multi-factor dynamic weighted algorithm on the experience signal values, professionalism signal values, authority signal values, and credibility signal values, for evaluating the target advertisement.

[0015] Optionally, the step of traversing all entity nodes and semantic relationship edges of the brand-specific knowledge graph, and labeling the successfully matched nodes and semantic relationship edges with corresponding signal levels and attaching verification metadata according to the preset EEAT authoritative signal rule library, includes: traversing all entity nodes and semantic relationship edges of the brand-specific knowledge graph using a depth-first search algorithm, extracting node attributes of each node, and extracting the relationship type of each edge; setting the EEAT authoritative signal rule library, including trigger conditions, signal types, signal levels, and verification metadata; matching the node attributes and relationship types with the trigger conditions in the EEAT authoritative signal rule library; and labeling and attaching verification metadata to the successfully matched nodes and semantic relationship edges according to the signal type and signal level, including: basic evidence information, verification status, and evidence chain ID.

[0016] Optionally, the overall authority index is: A=(α×E+β×Ex+γ×Au+δ×Tr)×CP Where A is the overall authority index, E is the empirical signal value, Ex is the professional signal value, Au is the authority signal value, Tr is the credibility signal value, C is the signal level correction coefficient, P is the negative information penalty factor, and α, β, γ, and δ are the weight coefficients of empirical signal, professional signal, authority signal, and credibility signal, respectively, and α+β+γ+δ=1.

[0017] On the other hand, this application also proposes a machine-readable storage medium storing instructions for causing a machine to perform the method for evaluating advertisements described above.

[0018] Through the aforementioned technical solution, this application constructs triplet data from multi-source heterogeneous brand data associated with advertising, and then fuses them to obtain a brand-specific knowledge graph, achieving a unified and structured expression of brand information. Based on four categories of rules—experience, professionalism, authority, and credibility—corresponding authoritative metadata and evidence chains are attached to graph nodes and relationships. This method deeply integrates the EEAT principle with the brand knowledge graph, achieving structured and standardized injection of authoritative signals, unlike traditional knowledge graphs which lack authoritative attribute annotations. This method solves the problems of AI source identification failure, fragmented brand authoritative information, and inability to be standardized and recognized by AI in a GEO environment, transforming brand assets into structured authoritative sources that machines can understand, thereby increasing the credibility and exposure weight of advertising content in AI-generated answers.

[0019] Other features and advantages of the embodiments of the present invention will be described in detail in the following detailed description section. Attached Figure Description

[0020] The accompanying drawings are provided to further illustrate embodiments of the present invention and form part of the specification. They are used together with the following detailed description to explain the embodiments of the present invention, but do not constitute a limitation thereof. In the drawings: Figure 1 This is a flowchart illustrating one method for evaluating advertisements according to this application; Figure 2 This is a schematic diagram of the brand knowledge graph construction and signal injection process in this application; Figure 3 This is a schematic diagram of the authoritative quantitative evaluation process for this application; Figure 4 This is a schematic diagram of an apparatus for evaluating advertisements according to this application.

[0021] Explanation of reference numerals in the attached figures 100 - Device for evaluating advertisements; 200 - Acquisition module; 300 - First processing module; 400 - Second processing module; 500 - Third processing module; 600 - Fourth processing module. Detailed Implementation

[0022] The specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are for illustration and explanation only and are not intended to limit the scope of the present invention.

[0023] It should be noted that the acquisition, transmission, storage, use, and processing of data in the technical solution of this application all comply with relevant laws and regulations. In the embodiments of this application, certain existing industry solutions such as software, components, and models may be mentioned. These should be considered exemplary, intended only to illustrate the feasibility of implementing the technical solution of this application, and do not imply that the applicant has already used or necessarily used such solutions.

[0024] This invention provides a method for evaluating advertisements, such as... Figure 1 As shown, the method includes: Step S101: Collect multi-source heterogeneous brand data of the target advertisement's associated brands, and perform preprocessing and structuring on the multi-source heterogeneous brand data to generate triplet data; Step S102: Based on the ontology customized for the brand marketing scenario, entity extraction, relation extraction and multi-dimensional knowledge fusion are performed on the triple data to construct a brand-specific knowledge graph. Step S103: Traverse all entity nodes and semantic relationship edges of the brand-specific knowledge graph, and according to the preset EEAT authoritative signal rule library, mark the corresponding signal level for the successfully matched nodes and semantic relationship edges and attach verification metadata. Step S104: Extract experience signal value, professionalism signal value, authority signal value and credibility signal value from the brand-specific knowledge graph based on the signal level and the verification metadata; Step S105: Calculate the overall authority index using a multi-factor dynamic weighting algorithm on the experience signal value, professionalism signal value, authority signal value, and credibility signal value, and use it to evaluate the target advertisement.

[0025] The aforementioned multi-source heterogeneous brand data includes data collected from official corporate data sources (such as official websites, annual reports, business registration information, trademark registration certificates, etc.), business data collected from internal corporate business systems (CRM, ERP, marketing management systems, etc.), and data collected from external authoritative data sources (such as industry association websites, authoritative media databases, award platforms, third-party rating agencies, etc.). The aforementioned preprocessing includes cleaning, deduplication, and formatting of the multi-source heterogeneous brand data. The aforementioned triplet data includes the subject, relationship, and attribute. The subject of the triplet data includes brand, founder, product, and industry organization; the relationship of the triplet data includes awards received, patents owned, media coverage by authoritative media, and professional qualifications; the attribute of the triplet data includes years of experience, certification level, number of cases, and official endorsement sources.

[0026] The ontology definition for the above-mentioned brand marketing scenario customization is as follows: the core entity types are defined as brand, founder, core product, core technology, and basic enterprise information; the core relationship types are defined as "create", "own", "produce", and "adopt"; the ontology definition includes extended entity types as customer cases, partners, sales channels, marketing activities, and supply chain nodes, and extended relationship types as "service", "cooperation", "sales", "hosting", and "supply"; the ontology definition also includes: defining authoritative entity types as awards, certifications, patents, media reports, industry standards, and third-party ratings; and defining authoritative relationship types as "obtain", "own", "be reported", "participate in formulation", and "be rated".

[0027] The aforementioned entity extraction includes extracting core entities and their attributes using rule matching and NLP techniques, such as brand name, establishment date, registered capital, founder's name, product model, and technical parameters; as well as extracting authoritative entities and their attributes, such as award name, issuing organization, award date, and award level; and patent information. The aforementioned relationship extraction involves extracting business entities and their detailed attributes, such as the industry, scale, cooperation period, and performance metrics of customer cases; and the time, location, number of participants, and conversion rate of marketing campaigns. The aforementioned brand-specific knowledge graph adopts a layered and progressive construction strategy, with the core layer as the foundation, the business layer as extensions, the authoritative layer as core features, and the scenario layer as application exit points. Data interoperability is achieved between layers through entity associations.

[0028] The aforementioned EEAT authoritative signal rule base transforms the judgment criteria of four types of signals—experience, professionalism, authority, and credibility—into machine-executable structured rules. Each rule includes: triggering conditions (entity type + relationship type + attribute value range), signal type, signal level (high / medium / low), and required metadata fields. The EEAT authoritative signal rule base incorporates industry-standard rule templates (e.g., patent signals for the technology industry, media reports for the FMCG industry) and supports enterprise-defined rule extensions. The EEAT authoritative signal rule base establishes a rule version management mechanism to record rule change history, ensuring the reproducibility of evaluation results. The aforementioned traversal of the brand-specific knowledge graph employs a depth-first search algorithm to traverse all entity nodes and semantic relationship edges of the brand-specific knowledge graph. For each node, its type, attribute key-value pairs, creation time, and update time are extracted; for each edge, its relationship type, starting node, ending node, and relationship weight are extracted. Nodes and edges that have already been labeled with signals are skipped to avoid redundant processing. Signal matching and preliminary identification include precise matching and fuzzy matching of the extracted node attributes and relationship types with the triggering conditions in the rule base. For example, precise matching triggers a professional signal if the entity type is "patent" and the relationship type is "ownership". Fuzzy matching uses a pre-trained NLP model to calculate the semantic similarity between entity names and attribute descriptions to identify potential authoritative signals that are not explicitly labeled (such as the authoritative signal corresponding to "industry leader").

[0029] The above signal level labeling method is based on a comprehensive evaluation of three dimensions: the authority of the signal source, the sufficiency of the evidence, and the timeliness. For example, high level: authoritative source at the national level or above, supported by more than 3 independent pieces of evidence, and published within 1 year; medium level: source at the provincial level or a leading industry player, supported by 1-2 independent pieces of evidence, and published within 1-3 years; low level: source at the local level or a general level, with only self-proving evidence, and published within 3-5 years. For multiple signals matched to the same entity or relationship, the highest level is taken as the final labeling result.

[0030] The aforementioned verification metadata is attached to each labeled signal, binding complete verification metadata, including: basic evidence information (source URL, publication time, publishing organization, author information), verification status (official link verification result, cross-platform consistency verification result), and evidence chain ID (a unique identifier for all evidence files associated with this signal). The metadata is stored in JSON-LD format and directly associated with knowledge graph nodes / edges. This method also includes content comparison of multiple sources for the same signal, calculating semantic similarity. If the similarity is ≥90%, it is marked as a "strong evidence chain," and the signal level is increased by one level; if the similarity is between 60% and 90%, it is marked as a "weak evidence chain," and the signal level remains unchanged; if the similarity is <60%, it is marked as a "conflicting evidence chain" and submitted for manual review.

[0031] Based on the brand-specific knowledge graph with annotations, standardized extraction of signal values ​​is achieved through multi-dimensional indicator aggregation and normalization. The specific process includes: signal dimension screening and data extraction, calculation of basic scores for single signals, summarization of scores for signals in the same dimension, industry benchmark normalization, and signal value verification and correction. Specifically, the above signal dimension screening and data extraction involves filtering the knowledge graph by signal type, extracting all nodes and edges labeled as experience, professionalism, authority, and credibility, and extracting basic indicators such as signal level, source level, evidence validity, timeliness, and consistency for each signal entry, while removing signal entries with invalid evidence (such as broken official links) or expired (publication time > 5 years). The above-mentioned single-signal basic score calculation calculates a basic score for each signal item using the following formula: Single-signal basic score = Signal level coefficient × Source level coefficient × Evidence validity coefficient × Timeliness coefficient. Signal level coefficient: High = 3, Medium = 2, Low = 1; Source level coefficient: International = 3, National = 2, Provincial / Industry = 1; Evidence validity coefficient: Strong evidence chain = 1, Weak evidence chain = 0.5, Self-evidence = 0.2; Timeliness coefficient: 1 - 0.1 × (Current year - Publication year), minimum 0.2. The above-mentioned signal scores within the same dimension are summarized by summing the basic scores of all single signals within the same dimension to obtain the original total score for that dimension. This includes deduplication of duplicate signals (the same event reported by multiple sources), retaining only the highest-scoring signal item, and truncating abnormally high values ​​to avoid individual extreme signals having an excessive impact on the overall result. This method introduces an industry benchmark database to obtain the average and highest signal scores of companies of similar size within the industry. The original total score is then converted into a normalized signal value between 0 and 1, using the formula: Normalized Signal Value = (Original Total Score - Lowest Industry Score) / (Highest Industry Score - Lowest Industry Score). If industry benchmark data is unavailable, the company's own historical highest score is used as the normalization upper limit. The calculated signal values ​​are then validated for reasonableness. If the deviation from the company's actual situation is too large, manual review is triggered. Based on the manual review results, the coefficients of individual signals are adjusted or erroneous signals are removed. The final empirical signal value, professional signal value, authoritative signal value, and credibility signal value are then output.

[0032] The aforementioned multi-factor dynamic weighting algorithm does not assign fixed weights to multiple influencing factors. Instead, it updates the weights in real time and periodically according to certain rules based on the factors' recent effectiveness, stability, or environmental adaptability, and then synthesizes the final decision score. The overall authority index is determined based on empirical signal values, professional signal values, authoritative signal values, credibility signal values, signal level correction coefficients, and negative information penalty factors.

[0033] Through the aforementioned technical solution, this application constructs triplet data from multi-source heterogeneous brand data associated with advertising, and then fuses them to obtain a brand-specific knowledge graph, achieving a unified and structured expression of brand information. Based on four categories of rules—experience, professionalism, authority, and credibility—corresponding authoritative metadata and evidence chains are attached to graph nodes and relationships. This method deeply integrates the EEAT principle with the brand knowledge graph, achieving structured and standardized injection of authoritative signals, unlike traditional knowledge graphs which lack authoritative attribute annotations. This method solves the problems of AI source identification failure, fragmented brand authoritative information, and inability to be standardized and recognized by AI in a GEO environment, transforming brand assets into structured authoritative sources that machines can understand, thereby increasing the credibility and exposure weight of advertising content in AI-generated answers.

[0034] In one embodiment, the step of traversing all entity nodes and semantic relationship edges of the brand-specific knowledge graph, and labeling the successfully matched nodes and semantic relationship edges with corresponding signal levels and attaching verification metadata according to a preset EEAT authoritative signal rule library, includes: traversing all entity nodes and semantic relationship edges of the brand-specific knowledge graph using a depth-first search algorithm, extracting node attributes of each node, and extracting the relationship type of each edge; setting the EEAT authoritative signal rule library, including trigger conditions, signal types, signal levels, and verification metadata; matching the node attributes and relationship types with the trigger conditions in the EEAT authoritative signal rule library; and labeling and attaching verification metadata to the successfully matched nodes and semantic relationship edges according to the signal type and signal level, including: basic evidence information, verification status, and evidence chain ID.

[0035] The four dimensions of EEAT are turned into scoring indicators and linked to each node of the knowledge graph. Among them, Experience is linked to real project cases, actual test use scenarios, and implementation results; Expertise is linked to qualification certificates, patents, team professional background, and years of in-depth industry experience; Authoritativeness is linked to official media reports, industry association endorsements, and third-party authoritative institution certifications; and Trustworthiness is linked to enterprise business information, compliance qualifications, real user reviews, and no record of false advertising.

[0036] Specifically, such as Figure 2 As shown, the aforementioned brand-specific knowledge graph adopts a layered and progressive construction strategy. The core layer is the foundation, the business layer is the extension, the authority layer is the core feature, and the scenario layer is the application exit. Data communication between layers is achieved through entity association. The specific construction method includes: Core Layer Construction: The core layer is the foundation of the knowledge graph, carrying the brand's most core identity information. Construction steps include: Ontology Definition: Defining core entity types as brand, founder, core products, core technologies, and basic company information; defining core relationship types as "established," "owned," "produced," and "adopted"; Data Collection: Collecting core data from official company data sources (website, annual reports, business registration information, trademark registration certificates) to ensure data authority and accuracy; Entity and Attribute Extraction: Using rule matching and NLP techniques to extract core entities and their attributes, such as brand name, establishment time, registered capital, founder's name, product model, and technical parameters; Entity Alignment and Disambiguation: Aligning different names of the same entity (e.g., brand abbreviation and full name, founder's former name); Disambiguating entities with the same name but different meanings (e.g., products with the same name from different companies), assigning a unique ID to each entity; Relationship Construction: Establishing basic relationships between core entities, forming the basic framework of the brand knowledge graph.

[0037] Business Layer Construction: The business layer expands upon the core layer with data on the brand's operational activities. Ontology Expansion: Entity types are expanded to include customer cases, partners, sales channels, marketing campaigns, and supply chain nodes; relationship types are expanded to include "service," "cooperation," "sales," "organizing," and "supply." Data Acquisition: Business data is collected from internal business systems (CRM, ERP, marketing management systems) and from public channels for information on partners and the supply chain. Entity and Relationship Extraction: Business entities and their detailed attributes are extracted, such as industry, scale, cooperation time, and performance metrics for customer cases; time, location, number of participants, and conversion rates for marketing campaigns. Connection to the Core Layer: Business layer entities are connected to core layer entities through "association" relationships, such as "customer case - association - brand" and "marketing campaign - promotion - product." Data Updates: A real-time data synchronization mechanism is established to automatically update the business layer knowledge graph when business system data changes.

[0038] Authority Layer Construction: This layer specifically carries the brand's authoritative information and EEAT signals. Authority Ontology Definition: Authority entity types are defined as awards, certifications, patents, media reports, industry standards, and third-party ratings; authority relationship types are defined as "obtained," "owned," "reported," "participated in formulation," and "rated." Multi-Source Authoritative Data Collection: Authoritative data is collected from external authoritative data sources (industry association websites, authoritative media databases, award platforms, and third-party rating agencies). Entity and Attribute Extraction: Authoritative entities and their attributes are extracted, such as award name, issuing institution, award time, and award level; patent application number, authorization time, inventor, and weighting, etc. Authority Relationship Construction: Relationships are established between authoritative entities and core / business layer entities, such as "brand-obtained-awards," "product-owned-patents," and "founder-reported-by-media." EEAT Signal Injection: All entities and relationships in the authority layer are labeled with EEAT signal levels and attached with verification metadata, forming a knowledge graph layer with authoritative attributes. Evidence Chain Management: An authoritative evidence chain database is established to store the original evidence files of all authoritative signals, supporting one-click traceability and verification. Scenario Layer Construction: The scenario layer aggregates and customizes underlying data for different brand marketing scenarios. Scenario Ontology Definition: Based on common brand marketing scenarios (product promotion, brand image building, crisis public relations, investment attraction, etc.), scenario-specific entity types and relationships are defined; Scenario-based Data Aggregation: For each specific scenario, relevant entities and relationships are filtered and aggregated from the core layer, business layer, and authority layer to form a scenario-specific sub-graph; Scenario Rule Configuration: Corresponding EEAT signal weights and evaluation rules are configured for each scenario, such as: Product Promotion Scenario: Professionalism signal weight 0.4, Experience signal weight 0.3, Authority signal weight 0.2, Credibility signal weight 0.1; Brand Image Building Scenario: Authority signal weight 0.4, Credibility signal weight 0.3, Experience signal weight 0.2, Professionalism signal weight 0.1; Scenario Output Interface: An independent API interface is provided for each scenario sub-graph, supporting calls from different application systems; Dynamic Optimization: Based on feedback on scenario application effects, the data range and rule configuration of the scenario sub-graph are dynamically adjusted to improve application effectiveness.

[0039] This method deeply integrates the EEAT principle with brand knowledge graphs, enabling the structured and standardized injection of authoritative signals, which is different from the shortcomings of traditional knowledge graphs that lack authoritative attribute labeling.

[0040] In one embodiment, the overall authority index is A = (α × E + β × Ex + γ × Au + δ × Tr) × CP, where A is the overall authority index, E is the empirical signal value, Ex is the professional signal value, Au is the authority signal value, Tr is the credibility signal value, C is the signal level correction coefficient, P is the negative information penalty factor, and α, β, γ, and δ are the weight coefficients of the empirical signal, professional signal, authority signal, and credibility signal, respectively, and α + β + γ + δ = 1.

[0041] The experience signals include linked customer cases, user empirical data, application timelines, and quantitative performance indicators; the professional signals include associated qualification certifications, patents, industry standards, and official verification links; the authoritative signals include marking the level of third-party authoritative sources, reporting time, scope of reprinting, and cross-platform consistency; and the credibility signals include integrating information timeliness, transparency, and consistency to generate basic authoritative labels.

[0042] Specifically, the experience signal value E reflects the brand's accumulation and effectiveness in practical applications. Quantitative indicators include: customer case study value, user validation value, application timeline value, and quantified effect indicator value. Specifically, the customer case study value = customer level coefficient × cooperation duration coefficient × effect quantification coefficient - customer level coefficient (e.g., Fortune 500 = 10, industry leader = 5, ordinary enterprise = 2) - cooperation duration coefficient: 1 - 0.1 × (current year - cooperation start year), minimum 0.2 - effect quantification coefficient: with clear ROI data = 1, with qualitative effect description = 0.5, without effect description = 0.2. The user validation value = user scale coefficient × validation type coefficient - user scale coefficient (e.g., 100,000+ users = 5, 10,000+ users = 2, 1,000+ users = 1) - validation type coefficient: third-party independent evaluation = 1, public user reviews = 0.5, brand self-certification = 0.2. The application timeline value = core product application years × 0.5, maximum 5 points. The above quantitative performance indicators are as follows: the score for each quantifiable performance indicator (such as conversion rate improvement or cost reduction) = indicator value × industry benchmark coefficient.

[0043] The above-mentioned original value of the empirical signal = all customer case values ​​+ all user empirical values ​​+ application timeline values ​​+ all quantitative performance indicator values. The above-mentioned empirical signal value = original value of the empirical signal / highest value of the industry empirical signal; Professionalism signal values ​​reflect a brand's expertise in technology and business. Quantitative indicators include: certification value, patent value, and industry standard value. Specifically, certification value = certification level coefficient × validity period coefficient × official verification coefficient - certification level coefficient (e.g., international certification = 10, national certification = 5, industry certification = 2); validity period coefficient: within validity period = 1, expired ≤ 1 year = 0.5, expired > 1 year = 0.2; official verification coefficient: with official verification link = 1, without = 0.3. The patent value = patent type coefficient × status coefficient × citation count coefficient - patent type coefficient (e.g., invention patent = 10, utility model patent = 5, design patent = 2); status coefficient: granted = 1, under substantive examination = 0.5, rejected / withdrawn = 0; citation count coefficient: citation count × 0.1, maximum 2 points. Industry standard value = Standard level coefficient × Participation role coefficient - Standard level coefficient, e.g., International standard = 10, National standard = 5, Industry standard = 2 - Participation role coefficient: Lead drafting = 1, Major participation = 0.5, General participation = 0.2. Original value of professional signal = All qualification certification values ​​+ All patent values ​​+ All industry standard values; Professional signal value = Original value of professional signal / Highest value of professional signal in the industry.

[0044] The aforementioned authoritative signal values ​​reflect a brand's influence and recognition within the industry. Quantitative indicators include: authoritative media coverage value, third-party institution rating value, and cross-platform consistency value. The authoritative media coverage value is calculated as follows: Authoritative Media Coverage Value = Media Level Coefficient × Reporting Time Coefficient × Reprint Scope Coefficient - Media Level Coefficient (e.g., Central-level media = 10, Provincial mainstream media = 5, Industry leading media = 2) - Reporting Time Coefficient: 1 - 0.05 × (Current Year - Reporting Year), minimum 0.2 - Reprint Scope Coefficient: Number of reprints × 0.01, maximum 2 points. The third-party institution rating value is calculated as follows: Rating Coefficient × Institutional Rating Coefficient - Rating Coefficient (e.g., AAA = 10, AA = 5, A = 2) - Institutional Rating Coefficient: International authoritative institutions = 1, Domestic authoritative institutions = 0.5. The cross-platform consistency value is calculated as follows: 5 points for the same information appearing on ≥3 independent authoritative platforms, 2 points for 2 platforms, and 0 points for 1 platform. The original authoritative signal value is calculated as follows: Total authoritative media coverage value + Total third-party institution rating value + Cross-platform consistency value. Authority signal value = Original authority signal value / Highest authority signal value in the industry; The aforementioned credibility signal values ​​reflect the authenticity and reliability of brand information. Quantitative indicators include: information timeliness, information transparency, information consistency, and negative information. Information timeliness is calculated as follows: 5 points for information published ≤ 1 year ago, 3 points for 1-3 years ago, 1 point for 3-5 years ago, and 0 points for more than 5 years ago. Information transparency is calculated as follows: 3 points for a clear source link, 1 point for a source name without a link, and 0 points for no source; 2 points for a complete chain of evidence, 1 point for a partial chain of evidence, and 0 points for no chain of evidence. Information consistency is calculated as follows: 2 points for complete consistency between internal and external public channels, 1 point for minor discrepancies, and 0 points for major contradictions. Negative information is calculated as follows: 2-10 points are deducted for each verified piece of negative information, determined based on its severity and scope of dissemination. The original credibility signal value is calculated as: Information Timeliness + Information Transparency + Information Consistency - Negative Information Value. The credibility signal value is calculated as: max(0, Original Credibility Signal Value / 10).

[0045] In one embodiment, the method further includes: the preprocessing includes cleaning, deduplication, and formatting of multi-source heterogeneous brand data. For example, data is collected from internal enterprise data sources (official website, CRM, white paper) and external public data sources (authoritative media, patent database, industry association, award platform), and then cleaned, deduplicated, and formatted to form a standard dataset.

[0046] This method is the first to deeply integrate the EEAT principle with brand knowledge graphs, enabling the structured and standardized injection of authoritative signals, unlike traditional knowledge graphs which lack authoritative attribute annotations. Furthermore, it constructs a multi-source evidence chain verification mechanism, ensuring the traceability and verifiability of authoritative signals through authoritative source binding, cross-platform consistency verification, and official link verification. This application forms a proactive authoritative source output solution for the GEO environment, allowing brand information to directly adapt to AI retrieval and generation logic. It employs an objective and quantifiable weighted calculation model, using the signal's inherent attributes as the calculation basis, eliminating subjective human scoring and ensuring the objectivity of the evaluation results.

[0047] According to one implementation method, such as Figure 3As shown, taking a certain technology company, A, as an example, the data collection process involves: The system automatically acquiring information from its official website, patent announcements, authoritative media reports, and Red Dot Design Award winners. Knowledge graph construction involves extracting entities such as Company A, AI chip A100, founder Zhang San, and the Red Dot Design Award, establishing relationships such as "production," "acquisition," and "award." Authoritative signal injection involves binding professional and authoritative signals and official verification links to the Red Dot Design Award; Xinhua News Agency reports are marked with high authority and credibility signals; and cross-platform information is consistently marked with high consistency signals. Authority calculation involves calculating a high authority index using a weighted formula. AI application involves prioritizing the brand information and generating authoritative citations when users query for reliable AI chips for industrial quality inspection. This method connects the authority index and the enhanced knowledge graph via API to AI question-answering systems, GEO optimization platforms, and brand asset management systems, providing authoritative sources for AI retrieval.

[0048] On the other hand, this application also proposes a device for evaluating advertisements, such as Figure 4 As shown, the device 100 for evaluating advertisements includes: an acquisition module 200, used to collect multi-source heterogeneous brand data associated with the target advertisement, and preprocess and structure the multi-source heterogeneous brand data to generate triple data; a first processing module 300, used to perform entity extraction, relation extraction and multi-dimensional knowledge fusion on the triple data based on an ontology customized for the brand marketing scenario, for constructing a brand-specific knowledge graph; a second processing module 400, used to traverse all entity nodes and semantic relationship edges of the brand-specific knowledge graph, and according to a preset EEAT authoritative signal rule library, mark the corresponding signal level for successfully matched nodes and semantic relationship edges and attach verification metadata; a third processing module 500, used to extract experience signal values, professionalism signal values, authority signal values ​​and credibility signal values ​​from the brand-specific knowledge graph according to the signal levels and the verification metadata; and a fourth processing module 600, used to calculate the overall authority index by performing a multi-factor dynamic weighted algorithm on the experience signal values, professionalism signal values, authority signal values ​​and credibility signal values, for evaluating the target advertisement.

[0049] In one embodiment, the second processing module is further configured to traverse all entity nodes and semantic relationship edges of the brand-specific knowledge graph using a depth-first search algorithm, extract node attributes of each node, and extract relationship types of each edge; set an EEAT authoritative signal rule base, including trigger conditions, signal types, signal levels, and verification metadata; match the node attributes and relationship types with the trigger conditions in the EEAT authoritative signal rule base; and attach verification metadata, including basic evidence information, verification status, and evidence chain ID, to the successfully matched nodes and semantic relationship edges according to the signal type and signal level.

[0050] Specifically, the overall authority index is: A = (α × E + β × Ex + γ × Au + δ × Tr) × CP, where A is the overall authority index, E is the empirical signal value, Ex is the professional signal value, Au is the authority signal value, Tr is the credibility signal value, C is the signal level correction coefficient, P is the negative information penalty factor, and α, β, γ, and δ are the weight coefficients of the empirical signal, professional signal, authority signal, and credibility signal, respectively, and α + β + γ + δ = 1.

[0051] Through the aforementioned technical solution, this application constructs triplet data from multi-source heterogeneous brand data associated with advertising, and then fuses them to obtain a brand-specific knowledge graph, achieving a unified and structured expression of brand information. Based on four categories of rules—experience, professionalism, authority, and credibility—corresponding authoritative metadata and evidence chains are attached to graph nodes and relationships. This method deeply integrates the EEAT principle with the brand knowledge graph, achieving structured and standardized injection of authoritative signals, unlike traditional knowledge graphs which lack authoritative attribute annotations. This method solves the problems of AI source identification failure, fragmented brand authoritative information, and inability to be standardized and recognized by AI in a GEO environment, transforming brand assets into structured authoritative sources that machines can understand, thereby increasing the credibility and exposure weight of advertising content in AI-generated answers.

[0052] The device 100 for evaluating advertisements includes a processor and a memory. The aforementioned acquisition module 200, first processing module 300, second processing module 400, third processing module 500, and fourth processing module 600 are all stored in the memory as program units. The processor executes the aforementioned program units stored in the memory to realize the corresponding functions.

[0053] The processor contains a kernel, which retrieves the corresponding program units from memory. One or more kernels can be configured; adjusting kernel parameters improves the accuracy of ad evaluation.

[0054] The memory may include non-permanent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM, and the memory includes at least one memory chip.

[0055] This invention provides a storage medium storing a program that, when executed by a processor, implements the method for evaluating advertisements.

[0056] This invention provides a processor for running a program, wherein the program executes the method for evaluating advertisements during runtime.

[0057] This invention provides a device including a processor, a memory, and a program stored in the memory and executable on the processor. When the processor executes the program, it performs the following steps: collecting multi-source heterogeneous brand data associated with a target advertisement; preprocessing and structuring the multi-source heterogeneous brand data to generate triplet data; performing entity extraction, relation extraction, and multi-dimensional knowledge fusion on the triplet data based on an ontology customized for the brand marketing scenario to construct a brand-specific knowledge graph; traversing all entity nodes and semantic relationship edges of the brand-specific knowledge graph, and according to a preset EEAT authoritative signal rule base, labeling successfully matched nodes and semantic relationship edges with corresponding signal levels and attaching verification metadata; extracting experience signal values, professionalism signal values, authority signal values, and credibility signal values ​​from the brand-specific knowledge graph based on the signal levels and verification metadata; and calculating an overall authority index using a multi-factor dynamic weighting algorithm on the experience signal values, professionalism signal values, authority signal values, and credibility signal values ​​to evaluate the target advertisement. The device in this document can be a server, PC, PAD, mobile phone, etc.

[0058] This application also provides a computer program product, which, when executed on a data processing device, is suitable for executing an initialization program with the following method steps: collecting multi-source heterogeneous brand data associated with the target advertisement; preprocessing and structuring the multi-source heterogeneous brand data to generate triple data; performing entity extraction, relation extraction, and multi-dimensional knowledge fusion on the triple data based on an ontology customized for the brand marketing scenario to construct a brand-specific knowledge graph; traversing all entity nodes and semantic relation edges of the brand-specific knowledge graph, and according to a preset EEAT authoritative signal rule base, labeling the successfully matched nodes and semantic relation edges with corresponding signal levels and attaching verification metadata; extracting experience signal values, professionalism signal values, authority signal values, and credibility signal values ​​from the brand-specific knowledge graph according to the signal levels and the verification metadata; and calculating the overall authority index by performing a multi-factor dynamic weighting algorithm on the experience signal values, professionalism signal values, authority signal values, and credibility signal values ​​to evaluate the target advertisement.

[0059] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0060] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0061] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0062] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0063] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0064] Memory may include non-persistent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0065] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0066] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0067] The above are merely embodiments of this application and are not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

Claims

1. A method of evaluating an advertisement, characterized by, The method includes: Collect multi-source heterogeneous brand data of brands associated with the target advertisement, and perform preprocessing and structuring on the multi-source heterogeneous brand data to generate triplet data; Based on the ontology customized for brand marketing scenarios, entity extraction, relation extraction, and multi-dimensional knowledge fusion are performed on the triplet data to construct a brand-specific knowledge graph. Traverse all entity nodes and semantic relationship edges of the brand-specific knowledge graph, and according to the preset EEAT authoritative signal rule library, mark the corresponding signal level for the successfully matched nodes and semantic relationship edges and attach verification metadata; Based on the signal level and the verification metadata, experience signal values, professionalism signal values, authority signal values, and credibility signal values ​​are extracted from the brand-specific knowledge graph. The overall authority index is calculated by a multi-factor dynamic weighting algorithm on the empirical signal value, professionalism signal value, authority signal value and credibility signal value, and is used to evaluate the target advertisement.

2. The method according to claim 1, characterized in that, The main components of the triplet data include brands, founders, products, and industry organizations; The relationships among the triplet data include awards received, patents owned, coverage by authoritative media, and professional qualifications. The attributes of the triplet data include years of experience, certification level, number of cases, and official endorsement sources.

3. The method of claim 1, wherein, The process involves traversing all entity nodes and semantic relationship edges of the brand-specific knowledge graph, and, based on the pre-defined EEAT authoritative signal rule base, labeling successfully matched nodes and semantic relationship edges with corresponding signal levels and attaching verification metadata, including: A depth-first search algorithm is used to traverse all entity nodes and semantic relationship edges of the brand-specific knowledge graph, extract the node attributes of each node, and extract the relationship type of each edge; Configure the EEAT authoritative signal rule base, including trigger conditions, signal type, signal level, and verification metadata; Match the node attributes and relationship types with the triggering conditions in the EEAT authoritative signal rule base; Based on the signal type and signal level, the nodes that are successfully matched and the semantic relationship edge are labeled with verification metadata, including: basic evidence information, verification status, and evidence chain ID.

4. The method according to claim 1, characterized in that, The overall authority index is: A=(α×E+β×Ex+γ×Au+δ×Tr)×CP Where A is the overall authority index, E is the empirical signal value, Ex is the professional signal value, Au is the authority signal value, Tr is the credibility signal value, C is the signal level correction coefficient, P is the negative information penalty factor, and α, β, γ, and δ are the weight coefficients of empirical signal, professional signal, authority signal, and credibility signal, respectively, and α+β+γ+δ=1.

5. The method according to claim 1 or 4, characterized in that, Empirical signal value = Original empirical signal value / Highest industry empirical signal value; Raw values ​​of empirical signals = values ​​of all customer case studies + values ​​of all user empirical evidence + values ​​of application timeline + values ​​of all quantitative performance indicators; Customer case value = Customer level coefficient × Cooperation duration coefficient × Quantitative effect coefficient - Customer level coefficient; Professional signal value = Original professional signal value / Highest professional signal value in this industry; The original value of a professional signal = the value of all qualification certifications + the value of all patents + the value of all industry standards; Certification value = Certification level coefficient × Validity period coefficient × Official verification coefficient - Certification level coefficient; Authority signal value = Original authority signal value / Highest authority signal value in the industry; The original value of authoritative signals = the value of all authoritative media reports + the value of all third-party ratings + the cross-platform consistency value; Confidence signal value = max(0, original confidence signal value / 10); The original value of the credibility signal = information timeliness value + information transparency value + information consistency value - negative information value.

6. The method according to claim 5, characterized in that, The empirical signals include linked customer cases, user empirical data, application timelines, and quantitative performance indicators; The professional signals include related qualification certifications, patents, industry standards, and official verification links; The authoritative signals include the level of the third-party authoritative source, the time of the report, the scope of reprinting, and cross-platform consistency; The credibility signal includes the timeliness, transparency, and consistency of the integrated information, generating a basic authoritative label.

7. The method of claim 1, wherein, The method also includes: The preprocessing includes cleaning, deduplication, and formatting of multi-source heterogeneous brand data.

8. An apparatus for evaluating an advertisement, characterized by The device includes: The acquisition module is used to collect multi-source heterogeneous brand data of the target advertisement's associated brands, and to preprocess and structure the multi-source heterogeneous brand data to generate triplet data. The first processing module is used to extract entities, extract relationships, and fuse multi-dimensional knowledge from the triplet data based on an ontology customized for brand marketing scenarios, in order to construct a brand-specific knowledge graph. The second processing module is used to traverse all entity nodes and semantic relationship edges of the brand-specific knowledge graph, and according to the preset EEAT authoritative signal rule library, to mark the corresponding signal level for the successfully matched nodes and semantic relationship edges and attach verification metadata. The third processing module is used to extract experience signal values, professionalism signal values, authority signal values ​​and credibility signal values ​​from the brand-specific knowledge graph based on the signal level and the verification metadata. The fourth processing module is used to calculate the overall authority index by performing a multi-factor dynamic weighting algorithm on the experience signal value, professionalism signal value, authority signal value and credibility signal value, which is used to evaluate the target advertisement.

9. The apparatus of claim 8, wherein, The process involves traversing all entity nodes and semantic relationship edges of the brand-specific knowledge graph, and, based on the pre-defined EEAT authoritative signal rule base, labeling successfully matched nodes and semantic relationship edges with corresponding signal levels and attaching verification metadata, including: A depth-first search algorithm is used to traverse all entity nodes and semantic relationship edges of the brand-specific knowledge graph, extract the node attributes of each node, and extract the relationship type of each edge; Configure the EEAT authoritative signal rule base, including trigger conditions, signal type, signal level, and verification metadata; Match the node attributes and relationship types with the triggering conditions in the EEAT authoritative signal rule base; Based on the signal type and signal level, the nodes that are successfully matched and the semantic relationship edge are labeled with verification metadata, including: basic evidence information, verification status, and evidence chain ID.

10. The apparatus according to claim 8, characterized in that, The overall authority index is: A=(α×E+β×Ex+γ×Au+δ×Tr)×CP Where A is the overall authority index, E is the empirical signal value, Ex is the professional signal value, Au is the authority signal value, Tr is the credibility signal value, C is the signal level correction coefficient, P is the negative information penalty factor, and α, β, γ, and δ are the weight coefficients of empirical signal, professional signal, authority signal, and credibility signal, respectively, and α+β+γ+δ=1.

11. A machine-readable storage medium having instructions stored thereon, the instructions comprising: This instruction is used to cause the machine to perform the method for evaluating advertisements as described in any one of claims 1-7 of this application.

Images