A Media Relationship Management Expert System

Abstract

The present invention discloses a media relationship management expert system. The system is divided into a three-layer architecture including a data access layer, a business logic layer and a presentation layer. The data access layer is used to provide a user interface to users, collect data in real time and pre-process the data to obtain multi-dimensional data; the business logic layer is constructed with a knowledge base, a rule base, an inference engine and a data analysis module that is arbitrarily combined by multiple mathematical models, and the data analysis module is used to analyze the multi-dimensional data to obtain data analysis results; the presentation layer is constructed with an online business module, and the online business module realizes the association of media relationships and the online business process of media relationship management based on the data analysis results and / or the results of logical reasoning operations. The media relationship management expert system realizes real-time dynamic detection and perception of changes in media relationships, and provides intelligent media social relationship management, which can significantly improve the quantity, scale, efficiency and professionalism of managing media relationships.

Description

Technical Field

[0001] The present invention relates to the technical field of media public relations management of public institutions or public figures, and in particular to a media relations management expert system. Background Art

[0002] Media relations and news communication management of public institutions or public figures is one of the core tasks of brand management. For institutions and individuals with brand management needs, the impact assessment generated by media reports is a prerequisite for conducting media social relationship management. The existing mainstream related technical solutions are still limited to public opinion monitoring systems and manually recorded media communication knowledge bases. The former collects media dissemination manuscripts from the entire network and provides suggestions for media public opinion handling, while the latter provides work guidance reference materials for professionals in this field, but neither provides dynamic digital management and intelligent expert-level management technical solutions. Summary of the invention

[0003] The present invention aims to at least solve the technical problems existing in the prior art. To this end, the present invention provides a media relationship management expert system, which realizes real-time dynamic detection and perception of changes in media relationships, and provides intelligent media social relationship management, which can significantly improve the number, scale, efficiency and professionalism of managing media relationships.

[0004] An embodiment of the present invention provides a media relationship management expert system, which is applied to a terminal device. The media relationship management expert system includes:

[0005] The data access layer is used to provide a user interface for user access, collect data in real time and pre-process the data to obtain multi-dimensional data;

[0006] The business logic layer is constructed with a knowledge base for storing media knowledge, a rule base for storing inference rules, an inference engine, and a data analysis module composed of a plurality of mathematical models in any combination, wherein the inference engine performs logical inference operations on undetermined media issues or media arguments based on the media knowledge and in accordance with the inference rules, and feeds back the results of the logical inference operations to the user through the data access layer; the data analysis module is used to analyze the multi-dimensional data to obtain data analysis results; the plurality of mathematical models include a communication influence evaluation model, a manuscript sentiment tendency analysis model, a manuscript text feature analysis model, a media and journalist relationship depth evaluation model, and a media tracking and reporting behavior feature model;

[0007] The presentation layer is constructed with an online business module, which is used to realize the association of media relations according to the data analysis results and / or the results of the logical reasoning operation, and realize the online business process of media relationship management.

[0008] According to the embodiments of the present invention, there are at least the following technical effects:

[0009] The system is constructed with a modular and process-based functional structure, and is divided into a three-layer architecture including a data access layer, a business logic layer, and a presentation layer. The data access layer is used to provide a user interface to users, collect data in real time, and pre-process the data to obtain multi-dimensional data; the business logic layer is constructed with a knowledge base, a rule base, an inference engine, and a data analysis module that is an arbitrary combination of multiple mathematical models, and the data analysis module is used to analyze multi-dimensional data to obtain data analysis results; the presentation layer is constructed with an online business module, and the online business module realizes the association of media relations and the online business process of media relationship management based on the data analysis results and / or the results of logical reasoning operations. The media relationship management expert system realizes real-time dynamic detection and perception of changes in media relations, and provides intelligent media social relationship management, which can significantly improve the quantity, scale, efficiency and professionalism of managing media relations.

[0010] Additional aspects and advantages of the present invention will be given in part in the following description and in part will be obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] The above and / or additional aspects and advantages of the present invention will become apparent and easily understood from the description of the embodiments in conjunction with the following drawings, in which:

[0012] Figure 1 A schematic diagram of the structure of a media relationship management expert system provided by one embodiment of the present invention;

[0013] Figure 2 A schematic diagram of a specific structure of an online business model provided by an embodiment of the present invention;

[0014] Figure 3 A schematic diagram of the specific structure of a data analysis module provided for one embodiment of the present invention. DETAILED DESCRIPTION

[0015] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and cannot be understood as limiting the present invention.

[0016] One embodiment of the present invention provides a media relationship management expert system, which is applied to terminal devices, where the terminal devices can be mobile electronic devices, non-mobile electronic devices and servers. Mobile electronic devices can be mobile phones, tablet computers, laptop computers, PDAs, vehicle-mounted electronic devices, wearable devices, super mobile personal computers, netbooks, personal digital assistants, etc.; non-mobile electronic devices can be personal computers, televisions, ATMs or self-service machines, etc.; the implementation scheme of the present invention is not specifically limited. The electronic device can include a processor, an external memory interface, an internal memory, a universal serial bus (USB) interface, a charging management module, a power management module, a battery, an antenna, a mobile communication module, a wireless communication module, an audio module, a speaker, a receiver, a microphone, a headphone interface, a sensor module, a button, a motor, an indicator, a camera, a display screen, and a subscriber identification module (SIM) card interface, etc. The server can be a stand-alone server or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content delivery networks (CDN), as well as big data and artificial intelligence platforms.

[0017] Reference Figures 1 to 3 Specifically, the media relationship management expert system provided in this embodiment includes a data access layer, a business logic layer and a presentation layer, wherein:

[0018] The data access layer is used to provide a user interface for user access, collect data in real time and pre-process the data to obtain multi-dimensional data.

[0019] The business logic layer is constructed with a knowledge base for storing media knowledge, a rule base for storing inference rules, an inference engine, and a data analysis module that is an arbitrary combination of multiple mathematical models. The inference engine performs logical inference operations on undetermined media issues or media arguments based on media knowledge and in accordance with inference rules, and feeds back the results of the logical inference operations to users through the data access layer; the data analysis module is used to analyze multi-dimensional data to obtain data analysis results; multiple mathematical models include a communication influence evaluation model, a manuscript sentiment tendency analysis model, a manuscript text feature analysis model, a media and journalist relationship in-depth evaluation model, and a media tracking and reporting behavior feature model.

[0020] The presentation layer is constructed with an online business module, which is used to realize the association of media relations based on the results of data analysis and / or logical reasoning operations, and to realize the online business process of media relationship management.

[0021] The system is constructed with a modular and process-based functional structure, and is divided into a three-layer architecture including a data access layer, a business logic layer, and a presentation layer. The data access layer is used to provide a user interface to users, collect data in real time, pre-process the data, obtain multi-dimensional data, and store the multi-dimensional data in a database; the business logic layer is constructed with a knowledge base, a rule base, an inference engine, and a data analysis module that is an arbitrary combination of multiple mathematical models, and the data analysis module is used to analyze the multi-dimensional data to obtain data analysis results; the presentation layer is constructed with an online business module, and the online business module realizes the association of media relations and the online business process of media relationship management based on the data analysis results and / or the results of logical reasoning operations. The media relationship management expert system realizes real-time dynamic detection and perception of changes in media relations, and provides intelligent media social relationship management, which can significantly improve the quantity, scale, efficiency and professionalism of managing media relations.

[0022] In some embodiments, the data access layer is used to collect data and pre-process the data to obtain multi-dimensional data, and the specific implementation is as follows:

[0023] The data access layer establishes a database after collecting data. For example, it collects media data from the entire network to establish a media library; obtains the latest releases of each media (i.e. media manuscripts) through real-time monitoring and establishes a manuscript library; performs data analysis and extraction on media manuscripts to obtain information on the media, media personnel, and institutions / persons mentioned in the manuscripts, thereby establishing an institution / person database and a reporter database, and at the same time expanding the media library.

[0024] Corresponding data is extracted from databases such as media libraries, manuscript libraries, organization / person libraries, and journalist libraries for preprocessing; for example, data such as media manuscripts, media, media personnel, organization / person information, etc. are associated, and then data aggregation, classification, clustering, natural language processing, etc. are performed to obtain multi-dimensional data.

[0025] As an optional implementation, the multi-dimensional data includes but is not limited to:

[0026] 1. Form: number of manuscripts, number of words, number of paragraphs, number of keywords (number of times the keywords of the organization / person appear in the manuscript);

[0027] 2. Influence: number of reposts, number of views, number of comments, number of likes, and interaction rate;

[0028] 3. Text characteristics: depth, liveliness, storytelling, and investigation;

[0029] 4. Media / media person attributes: field, line (sub-field), topic, tracking frequency, and emotional trend (friendly, supervisory, neutral).

[0030] The business logic layer first establishes a data relationship function library based on the actual business logic and multi-dimensional data of media relationship management, and then constructs a rule library based on the data relationship function library, for example: Rule Example 1, the higher the influence, the closer the relationship; Rule Example 2, the higher the investigation, the more attention it needs; Rule Example 3, the line focus is positively correlated with the stability of relationship maintenance; It should be noted that since there are enough dimensions to build relationships between various data, the rule library cannot be excluded from the protection scope of this method if it is not exhaustive in this embodiment. Finally, the index feature form is used to perform pattern matching with the rule library to build a knowledge base. In some embodiments, the knowledge base is through a media feature description and processing framework, an author feature description and processing framework, and a manuscript feature description and processing framework, where:

[0031] Media feature description and processing framework;

[0032] The data structure of the media feature description and processing framework includes: media name, media category, city, number of news, number of authors, areas of expertise, topic list, fan data, core data month-on-month, outstanding journalists, etc. The above data are formed into instance libraries and class libraries with feature labels through specific data modeling methods in different numerical relationships, forming a knowledge base.

[0033] Author characterization and processing framework;

[0034] The data structure of the author feature description and processing framework includes: author name, affiliation, field of expertise, topic distribution, number of manuscripts, line focus, topic concentration, manuscript depth, manuscript liveliness, manuscript storytelling, manuscript investigative nature, career, etc. Through specific data modeling methods, instance libraries and class libraries with feature labels are formed to form a knowledge base.

[0035] Manuscript feature description and processing framework;

[0036] The data structure of the manuscript feature description and processing framework includes: title, author, topic, word count, paragraph count, depth, liveliness, storytelling, investigative nature, number of readers, number of likes, number of comments, interaction rate, etc. Through specific data modeling methods, instance libraries and class libraries with feature labels are formed to form a knowledge base.

[0037] In some embodiments, the rule base includes:

[0038] Expert rule base for media outlets, for example: if the outlets are highly focused, it is recommended to prioritize long-term relationship maintenance; if the topic concentration is high, it is recommended to communicate only when the topic is appropriate; when seeking media coverage, give priority to reporters with high outlet focus, etc. Expert rule base for media level, for example: the higher the media influence index, the higher the weight of relationship maintenance. Expert rule base for topic quantity and distribution, for example: if the topic concentration is high, the accuracy of the topic selection is high, and if the topic distribution is wide, it is more willing to accept event invitations. Rule base for the division of labor of reporters from a single media, for example: if multiple reporters report on a specific company, it means that the division of labor between outlets is not strong; if a single reporter reports on a number higher than the second order of magnitude, it means that the division of labor between outlets is relatively concentrated but also free and unrestricted; if an object is repeatedly reported by a single reporter, it means that the division of labor between outlets is strict. In addition, there are expert rule bases on media tracking and reporting frequency, expert rule base on journalist tracking and reporting frequency, expert rule base on manuscript text characteristics, expert rule base on supervision and friendly tendencies, expert rule base on media regional competition impact, expert rule base on journalist career, expert rule base on institutional industry categories and attributes, expert rule base on institutional events and media coverage history, expert rule base on public policy trends affecting public relations strategies, and expert rule base on public opinion topics and corporate relevance strategies.

[0039] The business logic layer also constructs a data analysis module by arbitrarily combining multiple mathematical models, including a communication influence evaluation model, a manuscript sentiment analysis model, a manuscript text feature analysis model, a media and journalist relationship depth evaluation model, and a media tracking and reporting behavior feature model. The specific implementation of each mathematical model is as follows:

[0040] 1. Communication influence evaluation model;

[0041] 1. Select indicators such as total number of reads, highest number of reads, total number of likes + total number of comments, and number of words;

[0042] 2. Standardize the data of each indicator:

[0043] Suppose the selected k indicators are X1, X2, ..., X k , where X i ={X1,X2,...,X k}, n=1,7,30,... respectively correspond to the cycles of day, month, year, etc., and the value range is (0,1000). Assume that the cardinality of each indicator is: The standardized values ​​of each indicator data are Y1, Y2, ..., Y k , where Y i ={Y1,Y2,...,Y k},So:

[0044]

[0045] Set the upper limit of the standardized value, and finally Y i =(Y i >1000)1000:Y i ;

[0046] 3. Calculate the indicator weight based on information entropy:

[0047] According to the definition of information entropy in information theory, the information entropy of a set of data is:

[0048]

[0049] in If p i =0, then define

[0050] According to the calculation formula of information entropy, the information entropy of each indicator is calculated as E1, E2, ..., E k . The weight of each indicator is calculated by information entropy:

[0051]

[0052] 4. Find the influence index:

[0053] According to the standardized values ​​and weights of the above indicators, the index NI can be obtained:

[0054]

[0055] 2. Manuscript sentiment analysis model;

[0056] The emotional tendency of the manuscript is judged based on the title, which is divided into friendly, supervisory and neutral (that is, positive, negative and neutral).

[0057] 1. For example: Building a sentiment dictionary based on the open source "Tongyi Cilin" of Harbin Institute of Technology and Hownet of CNKI, and supplementing it with a negative word library and a degree word library;

[0058] 2. Manuscript title text segmentation. Perform word segmentation and part-of-speech tagging on the text based on the Jieba word segmentation database;

[0059] 3. Based on the sentiment word library, the sentiment of the segmented words is judged. When the number of negative words is odd, it is regarded as opposite polarity. The degree words are used to modify the sentiment intensity.

[0060] 4. Sentiment aggregation. The sentiment of a sentence is obtained by aggregating sentiment phrases. Suppose a sentence consists of n phrases:

[0061]

[0062] 3. Manuscript text feature analysis model;

[0063] The analysis of the text features of manuscripts is divided into story-telling, depth, serious or lively tendency, investigative, etc.

[0064] Story-telling model;

[0065] 1. Manually score the story-telling of a large number of manuscripts selected by sampling;

[0066] 2. Segment the text of the manuscripts based on the Jieba segmentation library, count the word frequencies and the manually annotated story-telling scores, and establish a story-telling word library. For example:

[0067] Taking the manual scoring results as the word weights, after processing a large number of articles, an evolving story-telling word library can be obtained. An example of the evolution is (taking the word "according to" as an example):

[0068]

[0069] Table 1

[0070] 3. Clean the word library. Clean and merge the word library according to the part of speech. Individual useless words can be removed, and low-frequency words can also be ignored. Words with little difference, such as time and place names, can be merged, etc.

[0071] 4. Based on the story-telling word library, segment and tokenize the text of the manuscripts and perform word weighted aggregation scoring.

[0072] Depth model:

[0073] Depth index H = N * Wn + S * Ws

[0074] N is the total number of personal names or personal pronouns in the manuscript, S is the length of the manuscript, and Wn and Ws are the weights corresponding to these two indicators respectively;

[0075] Serious or lively tendency model:

[0076] Referring to the story-telling word library, establish a serious or lively word library. Based on the serious word library and the lively word library, use the NLP model to score the seriousness and liveliness of each serious article and lively article; similar to the process of obtaining the quantization index in the story-telling dimension, the following lists a set of examples of obtaining the quantization index in the serious or lively tendency dimension: By segmenting words, filtering parts of speech and related processing, etc., establish serious and lively word libraries. The following only takes the serious word library as an example:

[0077]

[0078] Table 2

[0079] Based on the serious and lively word libraries, the news reports are scored for seriousness and liveliness respectively, and then the quantitative index of the serious or lively tendency dimension is calculated based on the two-pole scores. The quantitative index of the serious or lively tendency dimension is based on the seriousness score and the liveliness score. Here, a two-pole model is constructed based on these two scores, with the serious category as the left pole = 0 and the live category as the right pole = 1. The closer the index is to 0, the more it tends to be serious, and vice versa, the more it tends to be lively.

[0080] For example, the rating of a news report is:

[0081] Seriousness score = 1*9+1*7=16

[0082] Liveliness score = 1*8 = 8

[0083]

[0084] The index is close to 0, and the news report is a serious article.

[0085] Investigative Model:

[0086] Refer to the above story model and build an investigative vocabulary:

[0087] Survey index S = N*Wn+S*Ws+∑KW*Wkw

[0088] N is the total number of names or personal pronouns in the manuscript, S is the length of the manuscript, Wn and Ws are the weights corresponding to these two indicators, KW is the investigative word appearing in the manuscript, and Wkw is the weight of the corresponding word in the investigative vocabulary.

[0089] 4. In-depth evaluation model of the relationship between media and journalists;

[0090] For example, when counting the intervals between journalists' press releases, intervals of two months or more are considered as periods of cessation of cooperation, while intervals of less than two months are counted as continuous cooperation periods to obtain the final total cooperation period, which is then added with the number of press releases. The two calculation indicators are combined to generate a preliminary relationship depth index. In addition, the length of time since the most recent cooperation will also have a certain impact on the relationship depth. Therefore, the length of time since the most recent press release is added at the end to scale the index to obtain the final journalist relationship depth index. The same applies to the media.

[0091] Standardization of indicators:

[0092]

[0093]

[0094]

[0095] In some embodiments, the presentation layer is constructed with an online business module, which is used to realize the association of media relations according to the data analysis results and realize the online business process of media relationship management; wherein, the online business module includes a media reporting relationship discovery module, a communication influence evaluation module, a manuscript sentiment tendency analysis module, a manuscript text feature analysis module, a media and journalist relationship in-depth evaluation module, a media tracking and reporting behavior feature module, a media relationship maintenance decision-making suggestion module, a media social relationship intelligent assistance module and a peer competitor analysis module. The specific implementation of each module is as follows:

[0096] 1. The functions of the media coverage relationship discovery module are as follows:

[0097] The data access layer is used to collect data such as the number of readings, comments, likes, and reposts of the manuscript; the influence index of the manuscript is calculated using the communication influence evaluation model;

[0098] The influence index of the manuscript is associated with the media person and the media where the manuscript is located. The manuscripts are arranged according to the influence index to generate a list distributed by region, and then marked on the map according to the address of the media. The marking method can be any shape with a distinguishing color. Preferably, a radial halo display is used; the graphic size or color corresponding to the influence index is graded to form a radial heat map of the influence of brand organizations displayed at different levels;

[0099] Based on the relationship between manuscripts and authors and media, reports associated with brand organizations constitute graphic marks with different colors of media in the region. Colors and graphics are quantitatively superimposed according to the degree of regional concentration to establish a media concentration heat map, which vividly and intuitively shows the concentration of media with reporting relationships in different regions.

[0100] As an optional embodiment, by dragging the mouse to select on the media concentration heat map, a list of media within the geographical location range within the box can be obtained, and the list will display the number of media categories, the number of journalist categories, the number of media report categories, and the most influential articles, the most publishing journalists, and the most publishing media.

[0101] 2. The functions of the communication influence evaluation module are as follows:

[0102] Through the influence radiation heat map exposed by brand agencies, users are able to quickly and efficiently judge key media, and then select the corresponding media list to conduct influence assessment.

[0103] 3. The functions of the manuscript sentiment analysis module are as follows:

[0104] 1. Generate a comparative analysis data table of friendly reports and supervisory reports. The table displays the evaluation values ​​including the report strength value and ranking in the organization, the number of media that published the report, the cumulative number of topics, the global ranking of various values, and the reports with the highest friendly or supervisory strength.

[0105] 2. Generate a comparative radar chart for the statistical analysis of the topics of friendly reports and supervisory reports. The topics are extracted according to the keywords in the manuscript content and a set of valid selected words (topic keywords) are calculated using natural language processing methods. The characteristic descriptive words that effectively reflect the organization are sorted by word frequency and drawn into a gradient radar chart with the topic words as the corners of the graph.

[0106] 3. Generate a comparison table of classified subset data sorting (which means performing a round of index calculations in various categories, such as calculations for journalists in their media and calculations in different time intervals), including: rankings of friendliness or supervision intensity, rankings of the number of media reports, and rankings of the number of journalist reports. The ranking table shows the number of relevant indicators.

[0107] 4. Generate a pine needle comparison chart (deformed tower chart) of the timeline of friendly reports and supervisory reports, set different timeline intervals according to weeks, months, quarters, and years, and use pine needle column charts on both sides of the timeline to display different reporting influence indexes.

[0108] 5. Provide multiple dimensional analysis methods, preferably including comparative analysis based on geographical location, comparative analysis based on media line classification, comparative analysis based on influence index size list, etc.

[0109] 4. The functions of the manuscript text feature analysis module are as follows:

[0110] 1. The text feature analysis described here includes: depth, liveliness, storytelling, and investigation.

[0111] 2. The text feature analysis is statistically analyzed according to a certain time period. Preferably, the statistics are analyzed by week, month, quarter, and year, and the ranking is set according to various index categories.

[0112] 3. Text features are collected according to the names of journalists and media organizations, and the index rankings of journalists and media organizations are set separately.

[0113] 4. The subdivision category setting of text feature analysis can select more dimensions. Preferably, the indicators are counted according to the relevant manuscripts of a single reporter, the proportion of the time series interval of each indicator manuscript is compared, and the indicator ranking is compared with all the manuscripts. The statistics of each indicator on different topics are presented through radar charts; the subdivision category setting of the media refers to the method of reporters, and the ranking of subdivision indicators of different reporters is compared, and the comparison is made with quadrant interval diagrams.

[0114] 5. The functions of the in-depth assessment module of the relationship between media and journalists are as follows:

[0115] 1. Generate a chart including the time-series distribution of media and journalists’ press releases, a ranking of media relationship depth indicators, an analysis chart of media and journalists’ relationship depth indicators, and a concentration index chart of media tracking topics (for this organization / person); with journalists as the main body, the corresponding data and charts of the above four components are presented in a similar way.

[0116] 2. The media and journalists’ release time distribution chart uses the influence index as the vertical axis and the time interval as the horizontal axis. Each article is a scattered point, which eventually constitutes a scatter plot of all articles.

[0117] 3. The ranking of media relations depth indicators is displayed in a list including but not limited to progress bar chart, timestamp, average reporting frequency, average number of words (generated by algorithm), total number of reports, latest report titles, etc.

[0118] 4. The media and journalist relationship depth index analysis charts include but are not limited to the statistics of the number of reports of a single media or journalist, the trend curve diagram of the report frequency distribution, the report topic distribution ranking chart, the number of words ranking chart; the keyword distribution statistical features. The number of words is the statistical number of keywords appearing in the manuscript obtained by full-text matching according to the keyword set set by the organization.

[0119] 5. The media tracking topic concentration index chart includes the distribution of articles on the topic, the distribution over time, the ranking of the number of articles corresponding to the topic, and the ranking of the text feature index corresponding to the topic.

[0120] 6. The functions of the media and journalists’ tracking and reporting behavior feature module are as follows:

[0121] The components of this module include topic analysis charts, topic distribution charts, topic rankings, report-friendly and supervisory comparison statistics charts, the depth of relationships with reporters of a single media, reporter relationship depth rankings, internal reporter running line report distribution charts, internal reporter praise or supervisory report comparison analysis charts, media-related manuscript lists and query components. Based on all the aforementioned chart data, this system calculates and generates feature evaluation texts of media relations (a method of generating statements using data derivation). For example, fuzzy matching of some rules is used, such as data approximating a certain situation. This system calculates and generates feature evaluation texts of media relations, based on mathematical modeling of public relations business and setting dynamic parameters for basic analysis and simulation, combined with the logical judgment description of public relations business evolved by deep learning, to achieve high fitting indicators with real experts as much as possible.

[0122] 7. The functions of the media relations maintenance comparison chart module are as follows:

[0123] 1. The components of this module include but are not limited to: map search component, media display component, reporter information display component, full-dimensional query component, and marking component.

[0124] 2. The marking component is attribute-mapped with the reporter library and media library displayed by other database interface UIs.

[0125] 8. The functions of the media social relationship intelligent assistance module are as follows:

[0126] 1. This system calculates and generates feature evaluation texts of media relations to form a targeted strategy library. The strategy library provides basic logic for media relationship processing for different types of feature evaluations, and provides an interface for manual parameter adjustment to achieve hierarchical refinement.

[0127] 2. Based on the marking component of the above-mentioned media relationship maintenance comparison chart module, a recommendation component for media relationship processing is provided, a relationship processing recommendation list is provided for the media, and the recommended reason module, recommended indicator analysis scale, urgency, communication channel, and precautions are displayed in the list.

[0128] In some embodiments, the business logic layer is further constructed with a rule base and an inference engine for storing inference rules, wherein the inference engine performs logical inference operations on undetermined media issues or media arguments based on media knowledge and in accordance with inference rules, and the obtained logical inference operation results are fed back to the user through the data access layer; the data analysis module is used to analyze multi-dimensional data to obtain data analysis results. The purpose of the inference engine is to seek conclusions, solve problems and verify arguments.

[0129] In some embodiments, the business logic layer is also used to use feed-back deep learning to perform feedback optimization on the knowledge base and the inference engine, thereby achieving continuous optimization of the system and improving the efficiency and professionalism of the system.

[0130] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "illustrative embodiments", "examples", "specific examples", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0131] Although the embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the claims and their equivalents.

Claims

1. A media relations management expert system, It is characterized in that Applied to terminal equipment, the media relationship management expert system includes: The data access layer is used to provide a user interface for user access, collect data in real time and pre-process the data to obtain multi-dimensional data; the data access layer performs data association, data aggregation, data classification, data clustering and natural language processing on the data to obtain the multi-dimensional data; The multi-dimensional data includes: Morphological data, the morphological data at least includes: number of manuscripts, number of words, number of paragraphs, number of words; Influence data, which includes at least: number of reposts, number of reads, number of comments, number of likes, and interaction rate; Text feature data, the text feature data at least including: depth, liveliness, storytelling, and investigation; Media / media person attribute data, the media / media person attribute data at least including: field, line, topic, tracking frequency, and emotional trend; The business logic layer is constructed with a knowledge base for storing media knowledge, a rule base for storing inference rules, an inference engine, and a data analysis module composed of a plurality of mathematical models in any combination, wherein the inference engine performs logical inference operations on undetermined media issues or media arguments based on the media knowledge and in accordance with the inference rules, and feeds back the results of the logical inference operations to the user through the data access layer; the data analysis module is used to analyze the multi-dimensional data to obtain data analysis results; the plurality of mathematical models include a communication influence evaluation model, a manuscript sentiment tendency analysis model, a manuscript text feature analysis model, a media and journalist relationship depth evaluation model, and a media tracking and reporting behavior feature model; Among them, the in-depth evaluation model of the relationship between media and journalists includes: Standardization of indicators: ； ； ； The presentation layer is constructed with an online business module, which is used to realize the association of media relations according to the data analysis results and / or the results of the logical reasoning operation, and realize the online business process of media relationship management; The online business module at least includes a media coverage relationship discovery module, a manuscript sentiment analysis module, a media and journalist relationship depth assessment module, and a media tracking reporting behavior feature module; Among them, the media coverage relationship discovery module includes: The data access layer is used to collect the number of readings, comments, likes, and reposts of the manuscript; the influence index of the manuscript is calculated using the communication influence evaluation model; The influence index of the manuscript is associated with the media person and the media where the manuscript is located. The manuscripts are arranged according to their influence index to generate a list distributed by region. The relationship between the manuscript and the author and the media, and the reports associated with the brand organization constitute the graphic marks of the media in the region with different colors. The colors and graphics are quantitatively superimposed according to the regional concentration to establish the media concentration heat map; Among them, the manuscript sentiment analysis module includes: Generate a data table comparing friendly reports and supervisory reports. The table displays evaluation values ​​including the report strength value and ranking in the organization, the number of media that published the report, the cumulative number of topics, the global ranking of various values, and the reports with the highest friendly or supervisory strength; Generate a radar chart for the statistical analysis of friendly reports and supervisory reports. The topics are extracted from the keywords in the manuscript content and a set of valid words are calculated using natural language processing methods. The characteristic descriptive words that reflect the organization are sorted by word frequency and drawn into a gradient radar chart with the topic words as the corners of the graph; Generate a comparison table of classified subset data sorting, ranking of friendliness or supervision, ranking of number of media reports, ranking of number of journalist reports, and display the number of relevant indicators in the ranking table; Generate a pine needle comparison chart of the timeline of friendly reports and supervisory reports, set different timeline intervals according to weeks, months, seasons, and years, and use pine needle bar charts on both sides of the timeline to display different report influence indexes; Among them, the in-depth assessment module of the relationship between media and journalists includes: Generates media and journalists’ press release time distribution charts, media relationship depth index rankings, media and journalists’ relationship depth index analysis charts, and media tracking topic concentration index charts; Among them, the media and journalists’ tracking and reporting behavior feature modules include: Topic analysis charts, topic distribution charts, topic rankings, report-friendly and supervision comparison statistical charts, depth of relationships with reporters of a single media, ranking of reporter relationship depths, internal reporter running line report distribution charts, internal reporter praise or supervision report comparison analysis charts, media-related manuscript lists and query components.

2. A media relationship management expert system according to claim 1, It is characterized in that The data includes media data, media manuscripts, and media, media personnel, and organization / person information extracted from the media manuscripts.

3. A media relationship management expert system according to claim 1, It is characterized in that The business logic layer first establishes a data relationship function library according to the actual business logic of media relationship management and the multi-dimensional data, then constructs the rule library according to the data relationship function library, and finally performs pattern matching with the rule library in an exponential feature form to construct the knowledge base.

4. A media relationship management expert system according to claim 1, It is characterized in that The online business module implements the online business process of media relationship management through graphic components, table components, display components, query components, recommendation components and tag components.

5. A media relationship management expert system according to claim 1, It is characterized in that The rule base includes at least one of a media line expert rule base, a media level expert rule base, a topic quantity and distribution expert rule base, a single media reporter division of labor rule base, a media tracking and reporting frequency expert rule base, a reporter tracking and reporting frequency expert rule base, a manuscript text feature expert rule base, a supervision and friendly tendency expert rule base, a media regional competition impact expert rule base, a reporter career expert rule base, an institution industry category and attribute expert rule base, an institution event and media coverage history expert rule base, a public policy trend impact public relations strategy expert rule base, and a public opinion topic and enterprise relevance strategy expert rule base.

6. A media relationship management expert system according to claim 1, It is characterized in that The knowledge base is constructed based on the media feature description and processing framework, the author feature description and processing framework and the manuscript feature description and processing framework; wherein the data structure of the media feature description and processing framework includes the media name, media category, city, number of news, number of authors, areas of expertise, topic list, fan data, core data year-on-year growth and outstanding authors; the data structure of the author feature description and processing framework includes the author's name, unit, areas of expertise, topic distribution, number of manuscripts, line focus, topic concentration, manuscript depth, manuscript liveliness, manuscript storytelling, manuscript investigative nature and career; the data structure of the manuscript feature description and processing framework includes title, author, topic, word count, paragraph count, depth, liveliness, storytelling, investigative nature, reading volume, number of likes, number of comments and interaction rate.

7. A media relationship management expert system according to claim 1, It is characterized in that The business logic layer is also used to perform feedback optimization on the knowledge base using feed-back deep learning.

8. A media relationship management expert system according to claim 7, It is characterized in that The business logic layer is also used to perform feedback optimization on the inference engine using feedback deep learning.

Images