Enterprise content recommendation method and device, electronic equipment and storage medium

Abstract

The invention provides an enterprise content recommendation method and device, electronic equipment and a storage medium. According to the enterprise content recommendation method, firstly, enterprise information of all enterprises is obtained, and enterprise climate benefit portraits of all the enterprises are generated based on the enterprise information; secondly, for each enterprise, processing the enterprise climate benefit portrait of each enterprise through a preset recommendation algorithm, and generating an initial recommendation list of the current enterprise; and for each enterprise, matching the enterprise climate benefit portrait of the current enterprise with each enterprise climate benefit portrait in the initial recommendation list to generate a matching score corresponding to each enterprise climate benefit portrait in the initial recommendation list. And finally, for each enterprise, matching enterprise climate benefit portraits with scores meeting preset conditions from the initial recommendation list to generate a final recommendation list, and pushing the final recommendation list to the current enterprise.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to a method, device, electronic device and storage medium for enterprise content recommendation. Background Art

[0002] As an important part of green finance, the Climate Investment and Financing Platform aims to promote the formation of a systematic response between local policies and the realization of national climate change response and carbon peak and carbon neutrality goals through measures such as strengthening carbon accounting and information disclosure, model and tool innovation, and policy coordination. By building a national climate investment and financing project database, strengthening talent team building and international exchanges and cooperation, the Climate Investment and Financing Platform aims to provide financial institutions with better financial services, support the development of advanced low-carbon technologies, thereby promoting the green and low-carbon transformation of the whole society and helping to achieve the carbon peak and carbon neutrality goals.

[0003] On the existing platform, financial resources can only be guided to gather in the green and low-carbon fields, and only financial institutions are supported to connect with enterprises with climate benefits. The calculation of corporate climate benefits depends on evaluation by relevant departments. Therefore, content recommendations can only be made between financial institutions and enterprises, and content recommendations cannot be made between enterprises. Summary of the invention

[0004] In view of this, the present application provides a method, device, electronic device and storage medium for enterprise content recommendation to solve the problem that the existing technology can only recommend content between financial institutions and enterprises, but cannot achieve content recommendation between enterprises.

[0005] To achieve the above objectives, this application provides the following technical solutions:

[0006] The first aspect of the present application discloses a method for recommending enterprise content, comprising:

[0007] Obtaining enterprise information of each enterprise; wherein the enterprise information includes basic enterprise information, low-carbon project information, financial product information, green and low-carbon technology and policy document information;

[0008] Generate a corporate climate benefit portrait of each of the enterprises based on the enterprise information; wherein the corporate climate benefit portrait includes a project portrait, a financial product portrait, a technology portrait and a policy portrait;

[0009] For each enterprise, the enterprise climate benefit portraits of each enterprise are processed by a preset recommendation algorithm to generate an initial recommendation list for the current enterprise; wherein the initial recommendation list includes the enterprise climate benefit portraits recommended to the current enterprise;

[0010] For each enterprise, matching the climate benefit profile of the current enterprise with each climate benefit profile of the enterprise in the initial recommendation list, and generating a matching score corresponding to each climate benefit profile of the enterprise in the initial recommendation list;

[0011] For each enterprise, a final recommendation list is generated from the climate benefit portraits of enterprises whose matching scores meet preset conditions in the initial recommendation list, and the final push list is pushed to the current enterprise.

[0012] Optionally, the above method, after obtaining the enterprise information of each enterprise, further includes:

[0013] The enterprise information is cleansed to obtain cleansed data.

[0014] Optionally, in the above method, generating the enterprise climate benefit portrait of each enterprise based on the enterprise information includes:

[0015] Based on the enterprise information, generate labels for each of the enterprises under preset portrait dimensions; wherein the portrait dimensions include projects, financial products, technologies and policies;

[0016] The tags are refined and weighted by the TF-IDF algorithm to obtain refined tags; wherein the refined tags are accompanied by tag weights;

[0017] Performing weight balancing processing on the refined labels to obtain weight-balanced labels;

[0018] Calculating the attribute feature vector of each of the enterprise climate benefit portraits;

[0019] Based on the weight-balanced labels and the attribute feature vectors, a corporate climate benefit portrait of the enterprise is generated.

[0020] Optionally, the method described above, for each enterprise, matching the enterprise climate benefit portrait of the current enterprise with each enterprise climate benefit portrait in the initial recommendation list, and generating a matching score corresponding to each enterprise climate benefit portrait in the initial recommendation list, includes:

[0021] For each enterprise, calculate the similarity between the current enterprise's corporate climate benefit profile and the corporate climate benefit profiles of other enterprises;

[0022] Based on the label weights of the enterprise climate benefit portraits and the similarities, a matching score corresponding to each enterprise climate benefit portrait is generated.

[0023] Optionally, in the above method, for each enterprise, taking the climate benefit portrait of the enterprise whose matching score meets the preset conditions in the initial recommendation list as the recommendation content for the current enterprise includes:

[0024] Divide the climate benefit profiles of each enterprise in the initial recommendation list according to dimensions;

[0025] For each dimension of the corporate climate benefit profile, sort them from highest to lowest according to the matching score;

[0026] Filter out the corporate climate benefit portraits whose matching scores are in the top N and greater than a preset threshold in each dimension, generate a final recommendation list and push it to the current enterprise; wherein N is a preset positive integer.

[0027] The second aspect of the present application discloses a device for recommending enterprise content, comprising:

[0028] An acquisition unit is used to acquire enterprise information of each enterprise; wherein the enterprise information includes basic enterprise information, low-carbon project information, financial product information, green and low-carbon technology and policy document information;

[0029] A generating unit, configured to generate a corporate climate benefit portrait of each of the enterprises based on the enterprise information; wherein the corporate climate benefit portrait includes a project portrait, a financial product portrait, a technology portrait and a policy portrait;

[0030] A processing unit, for processing the enterprise climate benefit portrait of each enterprise through a preset recommendation algorithm, and generating an initial recommendation list for the current enterprise; wherein the initial recommendation list includes the enterprise climate benefit portrait recommended to the current enterprise;

[0031] A matching unit, for each enterprise, matches the climate benefit profile of the current enterprise with each climate benefit profile of the enterprise in the initial recommendation list, and generates a matching score corresponding to each climate benefit profile of the enterprise in the initial recommendation list;

[0032] The push unit is used to select the enterprise climate benefit portraits whose matching scores meet the preset conditions from the initial recommendation list for each enterprise to generate a final recommendation list, and push the final push list to the current enterprise.

[0033] Optionally, the above device further includes:

[0034] The data cleaning unit is used to clean the enterprise information to obtain cleaned data.

[0035] Optionally, in the above device, the generating unit includes:

[0036] A first generating subunit is used to generate labels for each of the enterprises under preset portrait dimensions based on the enterprise information; wherein the portrait dimensions include projects, financial products, technologies and policies;

[0037] A refining subunit, used to refine and weight the tags by using a TF-IDF algorithm to obtain refined tags; wherein the refined tags are accompanied by tag weights;

[0038] A processing subunit, used for performing weight balancing processing on the refined labels to obtain weight-balanced labels;

[0039] A first calculation subunit, used to calculate the attribute feature vector of each of the enterprise climate benefit portraits;

[0040] The second generating subunit is used to generate a corporate climate benefit portrait of the enterprise based on the weight-balanced label and the attribute feature vector.

[0041] Optionally, in the above device, the matching unit includes:

[0042] The second calculation subunit is used to calculate, for each enterprise, the similarity between the climate benefit portrait of the current enterprise and the climate benefit portraits of other enterprises;

[0043] The third generating subunit is used to generate a matching score corresponding to each enterprise climate benefit portrait based on the label weight and the similarity of the enterprise climate benefit portrait.

[0044] Optionally, in the above device, the pushing unit includes:

[0045] Divide sub-units to divide the climate benefit profiles of each enterprise in the initial recommendation list according to dimensions;

[0046] A sorting subunit is used to sort the corporate climate benefit portraits under each dimension from large to small according to the matching scores;

[0047] The screening subunit is used to screen out the corporate climate benefit portraits whose matching scores are in the top N and greater than a preset threshold in each dimension, generate a final recommendation list and push it to the current enterprise; wherein N is a preset positive integer.

[0048] The third aspect of the present application discloses an electronic device, comprising:

[0049] one or more processors;

[0050] a storage device having one or more programs stored thereon;

[0051] When the one or more programs are executed by the one or more processors, the one or more processors implement the method as described in any one of the first aspects of the present application.

[0052] A fourth aspect of the present application discloses a computer storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the method described in any one of the first aspects of the present application is implemented.

[0053] It can be seen from the above technical scheme that in a method for recommending enterprise content provided by this application, the enterprise information of each enterprise is first obtained, and the enterprise climate benefit portrait of each enterprise is generated based on the enterprise information. Then, for each enterprise, the enterprise climate benefit portrait of each enterprise is processed by a preset recommendation algorithm to generate an initial recommendation list for the current enterprise. For each enterprise, the enterprise climate benefit portrait of the current enterprise is matched with each enterprise climate benefit portrait in the initial recommendation list to generate a matching score corresponding to each enterprise climate benefit portrait in the initial recommendation list. Finally, for each enterprise, the enterprise climate benefit portraits whose matching scores meet the preset conditions are matched from the initial recommendation list to generate a final recommendation list, and the final push list is pushed to the current enterprise. It can be seen that by using the method of this application, by generating enterprise climate benefit portraits, the enterprise climate benefit portraits are matched between enterprises, and a recommendation list is generated for push. It not only supports mutual recommendations between financial institutions and enterprises, but also supports mutual recommendations between enterprises. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying any creative work.

[0055] Figure 1 A flowchart of a method for recommending enterprise content disclosed in an embodiment of the present application;

[0056] Figure 2 An example diagram of a project portrait disclosed in the embodiments of the present application;

[0057] Figure 3 This is an example diagram of the financial product portrait disclosed in the embodiments of this application;

[0058] Figure 4 An example diagram of the technical portrait disclosed in the embodiments of this application;

[0059] Figure 5 An example diagram of the policy portrait disclosed in the embodiments of this application;

[0060] Figure 6 This is a flowchart of an implementation of step S102 disclosed in an embodiment of the present application;

[0061] Figure 7 This is an example diagram of the financial product dimension features of the policy portrait disclosed in the embodiment of this application;

[0062] Figure 8 A matrix diagram formed according to the characteristic vector values ​​of each dimensional attribute of the policy portrait disclosed in the embodiment of the present application;

[0063] Fig. 9 A matching scoring diagram of the financial product portrait disclosed in the embodiment of this application;

[0064] Fig.10 A matching score diagram of the technical portrait disclosed in the embodiment of this application;

[0065] Fig.11 A matching scoring diagram of the policy portrait disclosed in the embodiment of this application;

[0066] Fig.12 This is a flowchart of an implementation of step S105 disclosed in an embodiment of the present application;

[0067] Fig.13 An example diagram of the corporate climate benefit portrait in each dimension disclosed in the embodiments of this application;

[0068] Fig.14 A schematic diagram of generating and pushing a final recommendation list disclosed in an embodiment of the present application;

[0069] Fig.15 A schematic diagram of a device for recommending enterprise content disclosed in an embodiment of the present application;

[0070] Fig.16 A schematic diagram of an electronic device disclosed in an embodiment of the present application. DETAILED DESCRIPTION

[0071] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0072] In this application, the terms "comprises", "comprising" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprising a ..." does not exclude the presence of other identical elements in the process, method, article or device comprising the element.

[0073] Furthermore, in this document, relational terms such as first and second, etc. are used only to distinguish one entity or operation from another entity or operation, but do not necessarily require or imply any such actual relationship or order between these entities or operations.

[0074] The present application embodiment provides a method for recommending enterprise content, such as Figure 1 As shown, specifically including:

[0075] S101. Obtain enterprise information of each enterprise; wherein the enterprise information includes basic enterprise information, low-carbon project information, financial product information, green and low-carbon technology and policy document information.

[0076] It should be noted that the enterprise information of each enterprise is first obtained, including basic enterprise information, low-carbon project information, financial product information, green and low-carbon technology and policy document information. Basic enterprise information includes enterprise name, business scope, registered capital, etc. Low-carbon project information includes project name, enterprise name, climate investment and financing level 1 category, keywords, etc. Financial product information includes product name, enterprise name, product term, applicable industry, etc. Green and low-carbon technology includes technology name, patent number, keywords. Policy document information includes policy name, issuing department, keywords, etc.

[0077] Optionally, in another embodiment of the present application, after executing step S101, the following may also be included:

[0078] Perform data cleansing on enterprise information to obtain cleaned data.

[0079] It should be noted that in order to remove redundant information and improve the subsequent data processing effect, after obtaining the enterprise information of each enterprise, the enterprise information is cleansed to obtain the cleaned enterprise information.

[0080] S102. Generate a climate benefit portrait of each enterprise based on enterprise information; wherein the climate benefit portrait includes project portrait, financial product portrait, technology portrait and policy portrait.

[0081] It should be noted that the collected enterprise information is processed through feature analysis and label extraction to generate a portrait based on the portrait generation rules, data range, portrait dimension, label score, scheduling time, etc. maintained by the administrator. Figure 2 As shown in Figure 2, the financial product profile is as follows: Figure 3 As shown, the technical portrait is as follows Figure 4 As shown in Figure 5 shown.

[0082] Optionally, in another embodiment of the present application, an implementation of step S102 is as follows: Figure 6 As shown, it may include:

[0083] S601. Generate labels for each enterprise under preset portrait dimensions based on enterprise information; wherein the portrait dimensions include projects, financial products, technologies and policies.

[0084] It should be noted that for each enterprise, the acquired enterprise information is sorted out to generate labels for each enterprise under preset dimensions; the dimensions include projects, financial products, technologies and policies.

[0085] S602: Refining and weighting the tags using the TF-IDF algorithm to obtain refined tags; wherein the refined tags are accompanied by tag weights.

[0086] It should be noted that the TF-IDF algorithm is used to refine the tags and set the weights to obtain the refined tags, wherein the refined tags are accompanied by tag weights. The TF-IDF algorithm is as follows:

[0087] For example, if there are 3 users and 4 tags, the relationship between tags and users will reflect the relationship between tags to a certain extent.

[0088] Let w(PT) represent the number of times a tag T is used to tag user P. TF(PT) represents the proportion of this tag count in all tags of user P, and the formula is as follows:

[0089] TF(P,T)=w(P,T) / ∑w(P,T)

[0090] Among them, w(P,T) is the number of a certain label attached to a user, and ∑w(P,T) is the total number of labels attached to the user.

[0091] For the above figure, user 1 has 5 A tags, 2 B tags, and 1 C tag, so the A tag TF of user 1 is 5 / (5+2+1).

[0092] The corresponding IDF(T) indicates the scarcity of tag T among all tags, that is, the probability of occurrence of this tag. If a tag 1 has a very small probability of occurrence and is used to tag a user at the same time, this makes the relationship between the user and the tag T closer.

[0093] IDF(P,T)=∑∑w(P,T) / ∑w(P,T)

[0094] Among them, ∑∑w(P,T) is the sum of all tags of all users. ∑w(P,T) is the sum of all tagged users.

[0095] Then, the weight value of the tag for the user can be obtained based on TF-IDF.

[0096] S603: Perform weight balancing processing on the refined labels to obtain weight-balanced labels.

[0097] It should be noted that after the initial label extraction, the label weight at this time only considers the relationship between the user and the label. In order to make the label more accurate, it is also necessary to consider business factors such as the behavior type and number of behaviors of the label. Therefore, the refined labels are weighted and balanced to obtain weight-balanced labels, making the refined labels more accurate.

[0098] Among them, weight balancing is simply to rebalance the weight of tag extraction based on business-related content such as user behavior type, behavior frequency, and time. The calculation logic is as follows:

[0099] The weight of user labels after balancing = behavior type weight * time decay * label weight of each user calculated by TF-IDF * number of behaviors.

[0100] Behavior type weight: Different behaviors such as user purchasing, usage, signing, browsing, clicking, and following have different importance and weights for different users.

[0101] Time decay: refers to the fact that the correlation between historical behavior and current behavior will continue to weaken over time. The time decay function assigns weights to user behavior data based on the order of occurrence. Some behaviors are affected by time and continue to weaken, so they should be multiplied by the time decay function. Some behaviors are not affected by time factors.

[0102] The tag weight of each user calculated by TF-IDF: the objective weight of each tag (this weight does not take into account factors such as behavior type, number of behaviors, and time).

[0103] Number of actions: The number of actions taken by a user also reflects the importance of the tag to him.

[0104] S604. Calculate the attribute feature vector of each enterprise's climate benefit portrait.

[0105] It should be noted that when constructing a portrait, the focus should be on clarifying the differences between the attribute characteristics of each dimension of the portrait. Therefore, after initially setting the dimensions and classifying the labels, it is necessary to quantify the attributes of each dimension to reflect the differences between the attributes of different feature dimensions, thereby supporting portrait clustering and forming a substantive basis for recommendation.

[0106] Taking the financial product portrait as an example, the portrait is modeled into 6 dimensions according to the actual business, and the characteristics of each dimension are as follows: Figure 7 As shown, each dimension contains the corresponding refined label.

[0107] In order to better display product features, distinguish the differences between features, and then support portrait clustering and form a substantial recommendation basis, the features of six dimensions are quantified, namely:

[0108] FPP={B,R,T,M,C,O}

[0109] The characteristic vector value of each dimension attribute is calculated through the SnowNLP library (a Python library for processing Chinese text), forming the following Figure 8 The matrix shown.

[0110] S605. Generate a corporate climate benefit portrait of the enterprise based on the weighted balanced labels and attribute feature vectors.

[0111] It should be noted that for portraits under different portrait dimensions, weight-balanced labels and attribute feature vectors are used for annotation to generate the corporate climate benefit portrait of the enterprise.

[0112] S103. For each enterprise, the enterprise climate benefit portrait of each enterprise is processed by a preset recommendation algorithm to generate an initial recommendation list for the current enterprise; wherein the initial recommendation list includes the enterprise climate benefit portrait recommended to the current enterprise.

[0113] It should be noted that for each enterprise, the enterprise climate benefit portrait of each enterprise is processed through a preset recommendation algorithm to generate an initial recommendation list for the current enterprise, wherein the initial recommendation list includes the enterprise climate benefit portrait recommended to the current enterprise, and the recommendation algorithm may adopt a collaborative filtering recommendation algorithm or a content-based recommendation algorithm. The collaborative filtering recommendation algorithm may adopt a product-based collaborative filtering algorithm or a user-based collaborative filtering algorithm.

[0114] The user-based collaborative filtering algorithm UserCF has a relatively simple idea. When a user A needs personalized recommendations, we can first find other users with similar interests to him, and then recommend products that those users like but have never heard of to A.

[0115] The two steps of the UserCF algorithm are:

[0116] 1. First, calculate the similarity between user A and other users based on the vector values ​​of each dimension of the user portrait (which have been calculated in the automatic annotation system), and find the N users most similar to A. The similarity algorithm uses cosine similarity for calculation, and the formula is as follows:

[0117]

[0118] 2. Make recommendations based on the product ratings given by these n users and their similarity with user A.

[0119] The basic idea of ​​item-based collaborative filtering (ItemCF) is to calculate the similarity between items based on the historical preference data of all users in advance, and then recommend items similar to the items that the user likes to the user. For example, items A and C are very similar, because users who like A also like C, so C is recommended to user A.

[0120] The ltemCF algorithm consists of two steps:

[0121] 1. Calculate the similarity between items. The similarity can be calculated using the Jaccard similarity coefficient.

[0122] J(A, B) = |A∩B| / |AUB|

[0123] As shown in the formula, the ratio of the number of elements in the intersection of two sets A and B to the number of elements in the union of A and B is called the Jaccard coefficient of the two sets, represented by the symbol J(A,B). The larger the Jaccard value, the higher the similarity.

[0124] 2. Generate a recommendation list for the user based on the similarity of items and the user’s historical behavior.

[0125] The essence of content-based recommendation algorithms is to analyze content and establish features; recommendations are made based on what features the user is interested in and what features a piece of content has. The specific process is as follows:

[0126] 1. Feature extraction

[0127] This process requires extracting the characteristics of users and products and the specific values ​​of quantitative representation. This part has been calculated when the portrait is generated. The main technologies used include TF-IDF-based weight algorithm, weight balance and quantitative representation calculation.

[0128] 2. User Preference Calculation

[0129] Utilize the user's past displayed ratings or implicit operation records to calculate the user's preference scores for different features (content attributes). The method for calculating preference scores can directly use statistical features, that is, calculate the user's scores under different labels. For example, user A's ratings for A1 type financial products are 4.0, 4.2, 4.4, and the ratings for A2 type financial products are 3.0, 2.8, 3.2. Then user A's score for A1 type financial products is 4.2 (average), and the score for A2 type financial products is 3.0. If it is a time-sensitive label feature, a time factor can be added. Calculation formula:

[0130] 4.0*αdiff1+4.2*adiff2+4.4*adiff3

[0131] Where a is a value less than 1, and diff is the time interval between the scores. The longer the interval, the lower the contribution of the score.

[0132] 3. Content recall

[0133] Match the features of the products to be recommended with the user's preference scores, and extract the pool of items that the user is most likely to like. For example, for user A in step 2, if the products to be recommended include some financial products of types A1 and A2, these should be included in user A's recommendation list.

[0134] S104. For each enterprise, the climate benefit portrait of the current enterprise is matched with each climate benefit portrait of the enterprise in the initial recommendation list, and a matching score corresponding to each climate benefit portrait of the enterprise in the initial recommendation list is generated.

[0135] It should be noted that for each enterprise, the current enterprise's climate benefit portrait is matched with each enterprise climate benefit portrait in the initial recommendation list, scored according to label weight and similarity, and a matching score corresponding to each enterprise climate benefit portrait in the initial recommendation list is generated.

[0136] Optionally, in another embodiment of the present application, an implementation of step S104 may include:

[0137] For each enterprise, the similarity between the climate benefit profile of the current enterprise and the climate benefit profiles of other enterprises is calculated.

[0138] Based on the label weights and similarities of the corporate climate benefit portraits, the matching scores corresponding to each corporate climate benefit portrait are generated.

[0139] It should be noted that for each enterprise, the cosine similarity is used to calculate the similarity between the current enterprise's corporate climate benefit portrait and the corporate climate benefit portraits of other enterprises. The formula is as follows:

[0140]

[0141] Then the label weight and similarity of the corporate climate benefit portrait are scored. The administrator can set an initial score for each label in advance according to the actual situation, multiply the initial score with the label weight and similarity, and generate the matching score corresponding to each corporate climate benefit portrait. Fig. 9 , Fig.10 and Fig.11 ,For example Fig. 9 For the Class A taxpayer in the project portrait and the Class A taxpayer in the financial product portrait, the initial score of the label is 10 points, the label weight is 1, and the similarity of the two labels is 100%, so the matching score is 10 points.

[0142] S105. For each enterprise, select the enterprise climate benefit portraits whose matching scores meet the preset conditions from the initial recommendation list to generate a final recommendation list, and push the final push list to the current enterprise.

[0143] It should be noted that in order to improve the accuracy of recommendations, for each enterprise, the corporate climate benefit portraits whose matching scores meet the preset conditions are screened from the initial recommendation list, and the screened contents are combined into a final recommendation list and recommended to the enterprise.

[0144] Optionally, in another embodiment of the present application, an implementation of step S105 is as follows: Fig.12 As shown, it may include:

[0145] S1201. Divide the climate benefit portrait of each enterprise in the initial recommendation list according to dimensions.

[0146] It should be noted that the climate benefit portraits of each enterprise in the initial recommendation list are first divided according to dimensions and sorted into a list of corporate climate benefit portraits under different portrait dimensions, such as Fig.13 As shown, the first table is a list of financial product portraits, the second table is a list of low-carbon technology portraits, and the third table is a list of policy portraits.

[0147] S1202. Sort the corporate climate benefit portraits in each dimension from large to small according to the matching scores.

[0148] It should be noted that see Fig.13 , for each dimension of corporate climate benefit portrait, sort them from large to small according to the matching score.

[0149] S1203. Filter out the corporate climate benefit portraits whose matching scores are in the top N and greater than a preset threshold in each dimension, generate a final recommendation list and push it to the current enterprise; where N is a preset positive integer.

[0150] It should be noted that in order to maintain the diversity of recommendations, we try to recommend relevant content in all dimensions, but at the same time we cannot recommend content with too low a matching score. Therefore, we select the corporate climate benefit portraits whose matching scores are in the top N and greater than the preset threshold in each dimension. For example, we select the corporate climate benefit portraits whose matching scores are in the top two and greater than 40 in each dimension, generate the final recommendation list, which can also be called a recommendation package, and push it to the current enterprise. Fig.14 shown.

[0151] In a method for recommending enterprise content provided in an embodiment of the present application, the enterprise information of each enterprise is first obtained, and the enterprise climate benefit portrait of each enterprise is generated based on the enterprise information. Then, for each enterprise, the enterprise climate benefit portrait of each enterprise is processed by a preset recommendation algorithm to generate an initial recommendation list for the current enterprise. Then, for each enterprise, the enterprise climate benefit portrait of the current enterprise is matched with each enterprise climate benefit portrait in the initial recommendation list, and a matching score corresponding to each enterprise climate benefit portrait in the initial recommendation list is generated. Finally, for each enterprise, a final recommendation list is generated from the enterprise climate benefit portraits in the initial recommendation list whose matching scores meet the preset conditions, and the final push list is pushed to the current enterprise. It can be seen that, using the method of the present application, by generating enterprise climate benefit portraits, the enterprise climate benefit portraits are matched between enterprises, and a recommendation list is generated for push. It not only supports mutual recommendations between financial institutions and enterprises, but also supports mutual recommendations between enterprises.

[0152] Another embodiment of the present application also provides a device for recommending enterprise content, such as Fig.15 As shown, including:

[0153] The acquisition unit 1501 is used to acquire the enterprise information of each enterprise; wherein the enterprise information includes basic enterprise information, low-carbon project information, financial product information, green and low-carbon technology and policy document information.

[0154] The generating unit 1502 is used to generate the enterprise climate benefit portrait of each of the enterprises based on the enterprise information; wherein the enterprise climate benefit portrait includes a project portrait, a financial product portrait, a technology portrait and a policy portrait.

[0155] Processing unit 1503 is used to process the corporate climate benefit portrait of each enterprise through a preset recommendation algorithm to generate an initial recommendation list for the current enterprise; wherein the initial recommendation list contains the corporate climate benefit portrait recommended to the current enterprise.

[0156] The matching unit 1504 is used to match the enterprise climate benefit portrait of the current enterprise with each enterprise climate benefit portrait in the initial recommendation list for each enterprise, and generate a matching score corresponding to each enterprise climate benefit portrait in the initial recommendation list.

[0157] The push unit 1505 is used to filter out the enterprise climate benefit portraits whose matching scores meet the preset conditions from the initial recommendation list for each enterprise to generate a final recommendation list, and push the final push list to the current enterprise.

[0158] In this embodiment, the specific execution process of the acquisition unit 1501, the generation unit 1502, the processing unit 1503, the matching unit 1504 and the push unit 1505 can be found in the corresponding Figure 1 The content of the method embodiment will not be repeated here.

[0159] In a device for recommending enterprise content provided by an embodiment of the present application, first, the acquisition unit 1501 acquires the enterprise information of each enterprise, and the generation unit 1502 generates the enterprise climate benefit portrait of each enterprise based on the enterprise information. Then, the processing unit 1503 processes the enterprise climate benefit portrait of each enterprise through a preset recommendation algorithm for each enterprise, and generates an initial recommendation list for the current enterprise. The matching unit 1504 then matches the enterprise climate benefit portrait of the current enterprise with each enterprise climate benefit portrait in the initial recommendation list for each enterprise, and generates a matching score corresponding to each enterprise climate benefit portrait in the initial recommendation list. Finally, the push unit 1505 generates a final recommendation list for each enterprise from the enterprise climate benefit portraits in the initial recommendation list whose matching scores meet the preset conditions, and pushes the final push list to the current enterprise. It can be seen that, by using the method of the present application, by generating enterprise climate benefit portraits, the enterprise climate benefit portraits are matched between enterprises, and a recommendation list is generated for push. It not only supports mutual recommendation between financial institutions and enterprises, but also supports mutual recommendation between enterprises.

[0160] Optionally, in another embodiment of the present application, the above-mentioned enterprise content recommendation device may further include:

[0161] The data cleaning unit is used to clean the enterprise information to obtain cleaned data.

[0162] In this embodiment, the specific execution process of the data cleaning unit can refer to the corresponding method embodiment described above, which will not be repeated here.

[0163] Optionally, in another embodiment of the present application, an implementation of the above-mentioned generating unit 1502 may include:

[0164] The first generating subunit is used to generate labels for each of the enterprises under preset portrait dimensions based on the enterprise information; wherein the portrait dimensions include projects, financial products, technologies and policies.

[0165] The refining subunit is used to refine and weight the tags by using the TF-IDF algorithm to obtain refined tags; wherein the refined tags are accompanied by tag weights.

[0166] The processing subunit is used to perform weight balancing processing on the refined labels to obtain weight-balanced labels.

[0167] The first calculation subunit is used to calculate the attribute feature vector of each of the climate benefit portraits of the enterprise.

[0168] The second generating subunit is used to generate a corporate climate benefit portrait of the enterprise based on the weight-balanced label and the attribute feature vector.

[0169] In this embodiment, the specific execution process of the first generating subunit, the refining subunit, the processing subunit, the first calculating subunit and the second generating subunit can be referred to in the corresponding Figure 6 The content of the method embodiment will not be repeated here.

[0170] Optionally, in another embodiment of the present application, an implementation of the matching unit 1504 may include:

[0171] The second calculation subunit is used to calculate, for each enterprise, the similarity between the climate benefit portrait of the current enterprise and the climate benefit portraits of other enterprises.

[0172] The third generating subunit is used to generate a matching score corresponding to each enterprise climate benefit portrait based on the label weight and the similarity of the enterprise climate benefit portrait.

[0173] In this embodiment, the specific execution process of the second calculation subunit and the third generation subunit can be found in the corresponding method embodiment described above, and will not be repeated here.

[0174] Optionally, in another embodiment of the present application, an implementation of the above-mentioned push unit 1505 may include:

[0175] The sub-units are used to divide the climate benefit portraits of each enterprise in the initial recommendation list according to dimensions.

[0176] The sorting subunit is used to sort the corporate climate benefit portraits under each dimension from large to small according to the matching scores.

[0177] The screening subunit is used to screen out the corporate climate benefit portraits whose matching scores are in the top N and greater than a preset threshold in each dimension, generate a final recommendation list and push it to the current enterprise; wherein N is a preset positive integer.

[0178] In this embodiment, the specific execution process of the division subunit, the sorting subunit and the screening subunit can be referred to in the corresponding Fig.12 The content of the method embodiment will not be repeated here.

[0179] Another embodiment of the application also provides an electronic device, such as Fig.16 As shown, specifically including:

[0180] One or more processors 1601.

[0181] The storage device 1602 stores one or more programs.

[0182] When one or more programs are executed by one or more processors 1601 , the one or more processors 1601 implement any one of the methods in the above embodiments.

[0183] Another embodiment of the present application further provides a computer storage medium on which a computer program is stored, wherein when the computer program is executed by a processor, any one of the methods in the above embodiments is implemented.

[0184] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can refer to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system or system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can refer to the partial description of the method embodiment. The system and system embodiments described above are merely schematic, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without creative work.

[0185] Professionals may further appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in the above description according to function. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians may use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0186] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for recommending enterprise content, characterized in that: include: Obtaining enterprise information of each enterprise; wherein the enterprise information includes basic enterprise information, low-carbon project information, financial product information, green and low-carbon technology and policy document information; Generate a corporate climate benefit portrait of each of the enterprises based on the enterprise information; wherein the corporate climate benefit portrait includes a project portrait, a financial product portrait, a technology portrait and a policy portrait; For each enterprise, the enterprise climate benefit portraits of each enterprise are processed by a preset recommendation algorithm to generate an initial recommendation list for the current enterprise; wherein the initial recommendation list includes the enterprise climate benefit portraits recommended to the current enterprise; For each enterprise, matching the climate benefit profile of the current enterprise with each climate benefit profile of the enterprise in the initial recommendation list, and generating a matching score corresponding to each climate benefit profile of the enterprise in the initial recommendation list; For each enterprise, a final recommendation list is generated from the climate benefit portraits of enterprises whose matching scores meet preset conditions in the initial recommendation list, and the final push list is pushed to the current enterprise.

2. The method according to claim 1, characterized in that After obtaining the enterprise information of each enterprise, the method further includes: The enterprise information is cleansed to obtain cleansed data.

3. The method according to claim 1, characterized in that Generating the enterprise climate benefit portrait of each enterprise based on the enterprise information includes: Based on the enterprise information, generate labels for each of the enterprises under preset portrait dimensions; wherein the portrait dimensions include projects, financial products, technologies and policies; The tags are refined and weighted by the TF-IDF algorithm to obtain refined tags; wherein the refined tags are accompanied by tag weights; Performing weight balancing processing on the refined labels to obtain weight-balanced labels; Calculating the attribute feature vector of each of the enterprise climate benefit portraits; Based on the weight-balanced labels and the attribute feature vectors, a corporate climate benefit portrait of the enterprise is generated.

4. The method according to claim 1, characterized in that: For each enterprise, matching the climate benefit profile of the current enterprise with each climate benefit profile of the enterprise in the initial recommendation list, and generating a matching score corresponding to each climate benefit profile of the enterprise in the initial recommendation list, includes: For each enterprise, calculate the similarity between the current enterprise's corporate climate benefit profile and the corporate climate benefit profiles of other enterprises; Based on the label weights of the enterprise climate benefit portraits and the similarities, a matching score corresponding to each enterprise climate benefit portrait is generated.

5. The method according to claim 1, characterized in that For each enterprise, the enterprise climate benefit portrait whose matching score meets the preset conditions in the initial recommendation list is used as the recommendation content of the current enterprise, including: Divide the climate benefit profiles of each enterprise in the initial recommendation list according to dimensions; For each dimension of the corporate climate benefit profile, sort them from highest to lowest according to the matching score; Filter out the corporate climate benefit portraits whose matching scores are in the top N and greater than a preset threshold in each dimension, generate a final recommendation list and push it to the current enterprise; wherein N is a preset positive integer.

6. A device for recommending enterprise content, characterized in that: include: An acquisition unit is used to acquire enterprise information of each enterprise; wherein the enterprise information includes basic enterprise information, low-carbon project information, financial product information, green and low-carbon technology and policy document information; A generating unit, configured to generate a corporate climate benefit portrait of each of the enterprises based on the enterprise information; wherein the corporate climate benefit portrait includes a project portrait, a financial product portrait, a technology portrait and a policy portrait; A processing unit, for processing the enterprise climate benefit portrait of each enterprise through a preset recommendation algorithm, and generating an initial recommendation list for the current enterprise; wherein the initial recommendation list includes the enterprise climate benefit portrait recommended to the current enterprise; A matching unit, for each enterprise, matches the climate benefit profile of the current enterprise with each climate benefit profile of the enterprise in the initial recommendation list, and generates a matching score corresponding to each climate benefit profile of the enterprise in the initial recommendation list; The push unit is used to select the enterprise climate benefit portraits whose matching scores meet the preset conditions from the initial recommendation list for each enterprise to generate a final recommendation list, and push the final push list to the current enterprise.

7. The device according to claim 6, characterized in that Also includes: The data cleaning unit is used to clean the enterprise information to obtain cleaned data.

8. The device according to claim 6, characterized in that The generating unit comprises: A first generating subunit is used to generate labels for each of the enterprises under preset portrait dimensions based on the enterprise information; wherein the portrait dimensions include projects, financial products, technologies and policies; A refining subunit, used to refine and weight the tags by using a TF-IDF algorithm to obtain refined tags; wherein the refined tags are accompanied by tag weights; A processing subunit, used for performing weight balancing processing on the refined labels to obtain weight-balanced labels; A first calculation subunit, used to calculate the attribute feature vector of each of the enterprise climate benefit portraits; The second generating subunit is used to generate a corporate climate benefit portrait of the enterprise based on the weight-balanced label and the attribute feature vector.

9. An electronic device, characterized in that: include: one or more processors; a storage device having one or more programs stored thereon; When the one or more programs are executed by the one or more processors, the one or more processors are enabled to implement the method according to any one of claims 1 to 5.

10. A computer storage medium, characterized in that: A computer program is stored thereon, wherein when the computer program is executed by a processor, the method according to any one of claims 1 to 5 is implemented.

Images