A digital integrated marketing service platform for meat products
By using data analysis, matching search units and reply processing units in the digital integrated marketing service platform for meat products, the optimal keywords are selected based on the number of keywords and the correlation degree, and combining pictures and user characteristics, the inability to be implemented in the existing technology is solved. The problem of meeting user needs improves the accuracy of reply text and data processing efficiency.
Patent Information
- Application Number
- CN202410228750.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-02-29
- Publication Date
- 2024-11-22
- Estimated Expiration
- 2044-02-29
AI Technical Summary
The prior art cannot effectively handle user problems that cannot be fully matched or multiple keywords exist, resulting in the reply text being unable to meet the actual needs of users.
A digital integrated marketing service platform for meat products was designed, including data analysis unit, matching search unit and reply processing unit. The keyword status is determined through keyword number and correlation degree, the optimal keyword is selected for text analysis, and combined with pictures Match status and user feature similarity determine the reply text.
It improves the accuracy of reply text and the accuracy of selection, meets user needs, and reduces data processing speed and keyword number.
Smart Images

Figure CN118037397B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of data processing, and in particular to a digital integrated marketing service platform for meat products. Background Art
[0002] Smart marketing is a new concept of innovative marketing that applies the integration of advanced computer, network, mobile Internet, Internet of Things and other scientific technologies to the field of contemporary brand marketing through human creativity, innovation and creative wisdom. Smart marketing includes technical means such as smart customer service, smart product recommendation and customer portrait analysis. The existing online smart marketing has the problem that automatic reply customer service cannot effectively understand user semantics, resulting in the inability to effectively meet customer needs. Therefore, how customer service can quickly and accurately understand user semantics to provide corresponding replies is an urgent problem to be solved by technical personnel in this field.
[0003] Chinese Patent Publication No. CN116842133A discloses a problem processing method, device, equipment and storage medium based on artificial intelligence, the method includes determining the first correlation between the problem to be processed and multiple standard answers in the answer library; determining the target answer from multiple standard answers according to the first correlation of multiple standard answers; and replying to the problem to be processed according to the target answer. In addition, Chinese Patent Publication No. CN113987151A discloses an intelligent customer service reply method and device, the method includes: if the answer to the question is not found, the answer to the question replied by the manual customer service is fed back to the user; the answer to the question replied by the manual customer service is semantically recognized, and the question and answer data is classified according to the question type and stored in the real-time batch knowledge base; the FAQ knowledge base is trained to obtain semantic knowledge using the SpanBERT model, and the base model is obtained by the knowledge distillation method; based on the base model, the weight of each type of part of speech is adjusted to obtain the adjusted TinyBERT model; the real-time batch knowledge base is trained using the TinyBERT model to obtain incremental question and answer knowledge, which is stored in the FAQ knowledge base. For questions that cannot be answered in the knowledge base, after the manual customer service replies, the question and answer data is recorded in a timely manner, and after training and sorting, it is updated to the knowledge base to achieve intelligent replies. It can be seen that the above technical solutions respectively disclose how to generate and classify knowledge bases with different answers and how to select the answer that meets the user's needs from multiple standard answers, but they do not consider the problem of how to select the reply text when the user question cannot be fully matched or there are multiple keywords in the user question, which may easily lead to the reply text failing to meet the actual needs of the user. Summary of the invention
[0004] To this end, the present invention provides a digital integrated marketing service platform for meat products, which is used to overcome the problem in the prior art that user questions cannot be fully matched or there are multiple keywords in the user questions, resulting in the inability to optimize the reply text and the reply text cannot meet the actual needs of the user.
[0005] To achieve the above objectives, the present invention provides a digital integrated marketing service platform for meat products, comprising:
[0006] A data analysis unit, for determining a keyword status according to the number of keywords in the target text of the target user and the keyword relevance, and determining, according to the keyword status, to select an optimal keyword for answering the text search according to the scene adaptability or to select an optimal keyword for answering the text search according to the keyword relevance;
[0007] A matching search unit connected to the data analysis unit is used to determine the optimal keyword according to the scene adaptability of each keyword, and to confirm the keyword relevance according to the number of periodic reflections or the number of associated combinations;
[0008] A reply processing unit connected to the data analysis unit and the matching search unit, for determining a target reply text according to a picture matching state or according to a user feature similarity when the number of reply texts is in a state where the number of reply texts is adjustable;
[0009] A pre-storage unit is connected to the data analysis unit, the matching search unit and the reply processing unit, and is used to store the purchase information of historical users, wherein the purchase information of a single historical user includes the purchased products, age, purchase quantity, number of repurchases, the historical user's corresponding question text and the historical reply text.
[0010] Furthermore, the data analysis unit determines a text analysis method according to the keyword status of the target text, including selecting the best keyword for answer text search according to scene adaptability, or selecting the best keyword for answer text search according to keyword association.
[0011] Furthermore, the data analysis unit determines the keyword status according to the keyword quantity and keyword association, including that the keyword quantity is equal to 1 or the keyword association is less than a preset keyword association, or that the keyword quantity is greater than 1 and the keyword association is greater than or equal to the preset keyword association.
[0012] Furthermore, the matching search unit calculates the scene adaptability of each keyword and determines the optimal keyword according to the scene adaptability of each keyword, including that when the scene adaptability of each keyword is less than a preset scene adaptability, the keyword with the largest scene adaptability is taken as the optimal keyword.
[0013] Or, if there are keywords whose scene adaptability is greater than or equal to a preset scene adaptability, all keywords whose scene adaptability is greater than or equal to the preset scene adaptability are recorded as optimal keywords.
[0014] Furthermore, the matching search unit records the keywords corresponding to the greatest keyword relevance as optimal keywords;
[0015] The matching search unit is provided with different confirmation modes of keyword relevance according to different numbers of keywords, including confirming the keyword relevance according to the number of periodic reflections, or confirming the keyword relevance according to the number of associated combinations.
[0016] Furthermore, the reply processing unit is also used to detect the number of reply texts, and, when the number of reply texts is in an adjustable reply text number state, determine the target reply text according to the image matching state, or determine the target reply text according to the user feature similarity.
[0017] Further, the reply processing unit detects whether there is a pre-storage defect in the goods in the target image in the image response state;
[0018] If there is a pre-stored defect, the reply processing unit records the reply text corresponding to the pre-stored defect as the target reply text;
[0019] If there is no pre-stored defect, the reply processing unit determines the target reply text according to the user feature similarity.
[0020] Further, the reply processing unit determines whether the target goods have pre-storage defects according to the packaging integrity and packaging color balance of the goods in the target image, and the pre-storage defects include first pre-storage defects and second pre-storage defects;
[0021] Among them, the first pre-stored defect is that the packaging integrity is within the defective integrity range, and the second pre-stored defect is that the packaging color balance is within the defective packaging color balance range.
[0022] Furthermore, the reply processing unit calculates the user feature similarity between the target user and each historical user corresponding to the most recent monitoring period, and records the historical user whose user feature similarity is closest to 1 as the comparison user. If the reply text corresponding to the target user contains the same reply text as that corresponding to the comparison user, then the text is used as the target reply text.
[0023] Furthermore, the user feature similarity is related to the number of times the user purchases the same product, age, purchase quantity, and number of repurchases.
[0024] Compared with the prior art, the beneficial effect of the present invention lies in that the technical solution of the present invention determines the keyword status according to the number of keywords and the keyword relevance, and selects different text analysis methods for different keyword statuses. In particular, for the case of multiple keywords, the text analysis method is determined according to the number and relevance of keywords, so that the selection of text analysis method is more in line with the actual work scenario, thereby improving the accuracy of the reply text of the present invention, thereby meeting the needs of users.
[0025] Furthermore, the matching search unit in the present invention calculates and determines the optimal keyword based on the scene adaptability of each keyword in turn, and selects keywords that cannot meet the actual scene through the scene adaptability, thereby making the selection of keywords more accurate and reducing the number of keywords to increase the data processing speed.
[0026] Furthermore, the reply processing unit in the present invention determines whether there are pre-storage defects based on the packaging integrity and packaging color balance of the goods in the target image, and selects a corresponding reply text when there are pre-storage defects, so that the selection of the reply text not only depends on the analysis results of the target text, but also combines the image analysis to make the selection of the reply text more accurate, thereby meeting the needs of users. BRIEF DESCRIPTION OF THE DRAWINGS
[0027] Figure 1 This is a unit connection diagram of the digital integrated marketing service platform for meat products according to an embodiment of the present invention;
[0028] Figure 2 This is a flow chart of an embodiment of the present invention for determining a text analysis method according to a keyword status;
[0029] Figure 3 The flowchart of an embodiment of the present invention is to determine the optimal keyword according to the scene adaptability of each keyword. DETAILED DESCRIPTION
[0030] In order to make the objects and advantages of the present invention more clearly understood, the present invention is further described below in conjunction with embodiments; it should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.
[0031] The preferred embodiments of the present invention are described below with reference to the accompanying drawings. It should be understood by those skilled in the art that these embodiments are only used to explain the technical principles of the present invention and are not intended to limit the protection scope of the present invention.
[0032] It should be noted that, in the description of the present invention, terms such as "up", "down", "left", "right", "inside" and "outside" indicating directions or positional relationships are based on the directions or positional relationships shown in the drawings. This is merely for the convenience of description and does not indicate or imply that the device or element must have a specific orientation, be constructed and operated in a specific orientation. Therefore, it cannot be understood as a limitation on the present invention.
[0033] In addition, it should be noted that in the description of the present invention, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For those skilled in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.
[0034] See also Figures 1 to 3 As shown, the present invention provides a digital integrated marketing service platform for meat products, comprising:
[0035] A data analysis unit, for determining a keyword status according to the number of keywords in the target text of the target user and the keyword relevance, and determining, according to the keyword status, to select an optimal keyword for answering the text search according to the scene adaptability or to select an optimal keyword for answering the text search according to the keyword relevance;
[0036] A matching search unit connected to the data analysis unit is used to determine the optimal keyword according to the scene adaptability of each keyword, and to confirm the keyword relevance according to the number of periodic reflections or the number of associated combinations;
[0037] A reply processing unit connected to the data analysis unit and the matching search unit, for determining a target reply text according to a picture matching state or according to a user feature similarity when the number of reply texts is in a state where the number of reply texts is adjustable;
[0038] A pre-storage unit is connected to the data analysis unit, the matching search unit and the reply processing unit, and is used to store the purchase information of historical users, wherein the purchase information of a single historical user includes the purchased products, age, purchase quantity, number of repurchases, the historical user's corresponding question text and the historical reply text.
[0039] The marketing service platform is connected to the user end and the manual end respectively. The target user uploads the target text through the user end, and the staff of the manual end can send manual reply information to the user end through the manual end.
[0040] The pre-storage unit further stores pre-storage goods images corresponding to each of the goods, first pre-storage defects, second pre-storage defects, and reply texts corresponding to the first pre-storage defects and the second pre-storage defects, respectively.
[0041] Specifically, the data analysis unit determines a text analysis method according to the keyword status of the target text, including selecting the best keyword for answering text search according to scene adaptability, or selecting the best keyword for answering text search according to keyword association.
[0042] Specifically, the data analysis unit determines the keyword status based on the number of keywords and the keyword association, the keyword status including a first preset keyword status in which the number of keywords is equal to 1 or the keyword association is less than a preset keyword association, and a second preset keyword status in which the number of keywords is greater than 1 and the keyword association is greater than or equal to the preset keyword association.
[0043] In the first preset keyword state, the text analysis method is to select the best keyword according to the scene adaptability to answer the text search;
[0044] In the second preset keyword state, the text analysis method is to select the best keyword to answer the text search based on the keyword relevance.
[0045] When the number of keywords is 0, the data analysis unit determines that the target text recognition has failed, and sends a reminder message for manual reply to the human end.
[0046] Specifically, the matching search unit calculates the scene adaptability of each keyword and determines the optimal keyword according to the scene adaptability of each keyword, including that the scene adaptability of each keyword is less than a preset scene adaptability, and the keyword with the largest scene adaptability is used as the optimal keyword.
[0047] Or, if there are keywords whose scene adaptability is greater than or equal to a preset scene adaptability, all keywords whose scene adaptability is greater than or equal to the preset scene adaptability are recorded as optimal keywords.
[0048] The calculation formula for the scene adaptation degree S of a single keyword is:
[0049]
[0050] Among them, p is the number of periodic reflections between the keyword and the target goods in the current recording period, p0 is the average number of periodic reflections of the keyword, p0 = the average value of the sum of the periodic reflections corresponding to the keyword and each recording period, the preset scene adaptation degree is 30% of p0, the periodic reflection number is the total number of occurrences of a single keyword for a single target product in the current recording period, that is, the total number of times all users use the keyword for a single target product in the current recording period, and the matching search unit is provided with a recording period, which is based on time, and every 15 days is recorded as a recording period.
[0051] Specifically, the matching search unit records the keywords corresponding to the greatest keyword relevance as the optimal keywords;
[0052] The matching search unit is provided with different keyword relevance confirmation modes for different keyword quantities, including confirming the keyword relevance according to the number of periodic reflections, or confirming the keyword relevance according to the number of associated matches.
[0053] If the number of keywords is less than the preset number of keywords, the keyword relevance is confirmed based on the number of periodic reflections;
[0054] If the number of keywords is greater than or equal to the preset number of keywords, the keyword relevance is confirmed based on the number of associated matches;
[0055] In confirming the keyword relevance based on the number of cycle reflections, the keyword relevance is positively correlated with the number of cycle reflections between the keyword and the target goods in the current recording cycle;
[0056] The keyword association degree is confirmed according to the number of associated collocations. The keyword association degree is positively correlated with the number of associated collocations. The number of associated collocations is confirmed by extracting all historical target texts corresponding to the current target goods in the current recording period, extracting two keywords in the current target text respectively, and detecting the number of times they appear together in the same historical target text, which is recorded as the keyword association degree. Any two keywords have corresponding keyword association degrees, and the minimum value of the keyword association degree is 0, that is, the number of times two keywords appear in the same historical target text is 0.
[0057] The preset keyword association degree is 10% of the total number of historical target texts in the current recording period. The value of the preset keyword number is related to the matching probability of each keyword association degree in the historical use process. The larger the matching probability, the more times different keywords in the historical target text are used in combination. Therefore, the matching probability of each keyword association degree in the historical use process is negatively correlated with the value of the preset keyword number. A value of the preset keyword number is provided, and the preset keyword number is 6.
[0058] Specifically, the reply processing unit detects the number of reply texts, and when the number of reply texts is in an adjustable reply text number state, determines the target reply text according to the image matching state, or determines the target reply text according to the user feature similarity.
[0059] In the present invention, each keyword is provided with a corresponding number of reply texts, which are stored in a pre-storage unit. The number of reply texts is the number of reply texts matched according to the optimal keyword. The adjustable reply text number state is that the reply text number is greater than the preset reply text number, and the preset reply text number is the optimal keyword number.
[0060] Specifically, the reply processing unit detects whether there is a pre-storage defect in the goods in the target image in the image response state;
[0061] If there is a pre-stored defect, the reply processing unit records the reply text corresponding to the pre-stored defect as the target reply text;
[0062] If there is no pre-stored defect, the reply processing unit determines the target reply text according to the user feature similarity.
[0063] The picture response state is that the target picture exists and the number of reply texts is in an adjustable reply text number state, and the target picture is sent by the user to the information receiving end;
[0064] The image matching status is that the goods in the target image have a pre-stored defect target, or the goods in the target image do not have a pre-stored defect target.
[0065] Specifically, the reply processing unit determines whether the target goods have pre-storage defects according to the packaging integrity and packaging color balance of the goods in the target image, and the pre-storage defects include first pre-storage defects and second pre-storage defects;
[0066] Among them, the first pre-stored defect is that the packaging integrity is within the defective integrity range, and the second pre-stored defect is that the packaging color balance is within the defective packaging color balance range.
[0067] The packaging integrity is the similarity of the contours of the goods in the target image and the pre-stored goods image. For the target image, the contour of the goods in the target image is first extracted by image analysis software, and the volume corresponding to the goods is estimated according to the genetic learning model. The contour similarity is the absolute value of the volume difference between the goods in the target image and the goods in the pre-stored goods image. The values within the preset defect integrity range are all greater than 20% of the volume of the pre-stored goods image.
[0068] The packaging color balance is the maximum color difference reference value of the goods in the target image, that is, the absolute value of the difference between the area corresponding to the maximum color chromaticity of the goods surface in the target image and the area corresponding to the minimum color chromaticity of the goods surface. The values within the range of defective packaging color balance are all less than 30% of the area corresponding to the maximum color chromaticity of the goods surface. The color chromaticity of the goods in the target image can be detected by colorimetric analysis software.
[0069] Specifically, if the target goods do not have the first pre-stored defect and the second pre-stored defect or do not have the target picture and the number of reply texts is in an adjustable reply text number state, the reply processing unit calculates the user feature similarity between the target user and each historical user corresponding to the most recent monitoring period, and records the historical user whose user feature similarity is closest to 1 as the comparison user. If the reply text corresponding to the target user contains the same reply text as that corresponding to the comparison user, then the text is used as the target reply text.
[0070] Specifically, the user feature similarity is related to the number of times the user purchases the same product, age, purchase quantity, and number of repurchases.
[0071] The user feature similarity is denoted as Y, and Y is set to be (y1 / y10)×α1+(y2 / y20)×α2+(y3 / y30)×α3+(y4 / y40)×α4;
[0072] Among them, y1 is the number of times the comparison user purchases the same product, y2 is the age of the comparison user, y3 is the purchase quantity of the comparison user, y4 is the number of repurchases, y10 is the number of times the target user purchases the same product, y20 is the age of the target user, y30 is the purchase quantity of the target user, y40 is the number of repurchases of the target user, α1=α2=α3=α4=0.25.
[0073] So far, the technical solutions of the present invention have been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it is easy for those skilled in the art to understand that the protection scope of the present invention is obviously not limited to these specific embodiments. Without departing from the principle of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will fall within the protection scope of the present invention.
[0074] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.
Claims
1. A digital integrated marketing service platform for meat products, characterized in that: include: A data analysis unit, for determining a keyword status according to the number of keywords in the target text of the target user and the keyword relevance, and determining, according to the keyword status, to select an optimal keyword for answering the text search according to the scene adaptability or to select an optimal keyword for answering the text search according to the keyword relevance; A matching search unit connected to the data analysis unit is used to determine the optimal keyword according to the scene adaptability of each keyword, and to confirm the keyword relevance according to the number of periodic reflections or the number of associated combinations; A reply processing unit connected to the data analysis unit and the matching search unit, for determining a target reply text according to a picture matching state or according to a user feature similarity when the number of reply texts is in a state where the number of reply texts is adjustable; A pre-storage unit connected to the data analysis unit, the matching search unit and the reply processing unit, for storing the purchase information of historical users, wherein the purchase information of a single historical user includes the purchased products, age, purchase quantity, number of repurchases, question texts corresponding to the historical user and historical reply texts of the historical user; The matching search unit calculates the scene adaptability of each keyword and determines the optimal keyword according to the scene adaptability of each keyword, including that the scene adaptability of each keyword is less than the preset scene adaptability, and the keyword with the largest scene adaptability is taken as the optimal keyword, Or, if there are keywords whose scene adaptability is greater than or equal to a preset scene adaptability, all keywords whose scene adaptability is greater than or equal to the preset scene adaptability are recorded as optimal keywords; The matching search unit is provided with different keyword relevance confirmation modes for different keyword quantities, including confirming the keyword relevance according to the number of periodic reflections, or confirming the keyword relevance according to the number of associated matches. If the number of keywords is less than the preset number of keywords, the keyword relevance is confirmed based on the number of periodic reflections; If the number of keywords is greater than or equal to the preset number of keywords, the keyword relevance is confirmed based on the number of associated matches; In confirming the keyword relevance based on the number of cycle reflections, the keyword relevance is positively correlated with the number of cycle reflections between the keyword and the target goods in the current recording cycle; The keyword association degree is confirmed according to the number of association collocations. The keyword association degree is positively correlated with the number of association collocations. The number of association collocations is confirmed by extracting all the historical target texts corresponding to the current target goods in the current recording period, extracting two keywords in the current target text respectively, and detecting the number of times they appear together in the same historical target text, which is recorded as the keyword association degree. Any two keywords have corresponding keyword association degrees, and the minimum value of the keyword association degree is 0, that is, the number of times two keywords appear in the same historical target text is 0; The reply processing unit detects whether there is a pre-storage defect in the goods in the target image in the image response state; If there is a pre-stored defect, the reply processing unit records the reply text corresponding to the pre-stored defect as the target reply text; If there is no pre-stored defect, the reply processing unit determines a target reply text according to the user feature similarity; The reply processing unit determines whether the target goods have pre-storage defects according to the packaging integrity and packaging color balance of the goods in the target image, and the pre-storage defects include first pre-storage defects and second pre-storage defects; The first pre-stored defect is that the package integrity is within the defect integrity range, and the second pre-stored defect is that the package color balance is within the defect package color balance range; The reply processing unit calculates the user feature similarity between the target user and each historical user corresponding to the most recent monitoring period, and records the historical user whose user feature similarity is closest to 1 as the comparison user. If the reply text corresponding to the target user contains the same reply text as that corresponding to the comparison user, then the text is used as the target reply text.
2. The digital integrated marketing service platform for meat products according to claim 1, characterized in that: The data analysis unit determines a text analysis method according to the keyword status of the target text, including selecting the best keyword for answering text search according to scene adaptability, or selecting the best keyword for answering text search according to keyword association.
3. The digital integrated marketing service platform for meat products according to claim 2, characterized in that: The data analysis unit determines the keyword status according to the number of keywords and the keyword association, including that the number of keywords is equal to 1 or the keyword association is less than a preset keyword association, or that the number of keywords is greater than 1 and the keyword association is greater than or equal to the preset keyword association.
4. The digital integrated marketing service platform for meat products according to claim 3, characterized in that: The matching search unit records the keywords corresponding to the greatest keyword relevance as the optimal keywords; The matching search unit is provided with different confirmation modes of keyword relevance according to different numbers of keywords, including confirming the keyword relevance according to the number of periodic reflections, or confirming the keyword relevance according to the number of associated combinations.
5. The digital integrated marketing service platform for meat products according to claim 4, characterized in that: The reply processing unit is also used to detect the number of reply texts, and, when the number of reply texts is in an adjustable reply text number state, determine the target reply text according to the image matching state, or determine the target reply text according to the user feature similarity.
6. The digital integrated marketing service platform for meat products according to claim 5, characterized in that: The user feature similarity is related to the number of times the user purchases the same product, age, purchase quantity, and number of repurchases.
