Building business opportunity recommendation method and device, computer equipment and storage medium

Abstract

The invention relates to the technical field of data processing, and discloses a building business opportunity recommendation method and device, computer equipment and a storage medium, and the method comprises the steps: obtaining historical business opportunity data of a historical engineering project, and obtaining bidding data of a bidding object; associating the historical business opportunity data with the bidding data to generate original data; preprocessing the original data, setting a corresponding label, and generating sample data with the label; performing model training according to the sample data to generate a business opportunity recommendation model; the business opportunity recommendation model is used for recommending corresponding business opportunity data to the target object. According to the business opportunity recommendation method and device, multi-dimensional structured original data can be formed, model training is subsequently carried out based on the original data, the model can learn features related to an engineering project, the business opportunity recommendation model obtained through training can recommend business opportunities in the building field, and the business opportunity recommendation efficiency is improved. And the efficiency of acquiring high-quality business opportunity data by business personnel is greatly improved.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a method, device, computer equipment and storage medium for recommending construction business opportunities. Background Art

[0002] In the business opportunity recommendation scenario, faced with massive amounts of business opportunity data, it is necessary to filter out the business opportunity information that matches specific users and push it to users to help them select business opportunities efficiently.

[0003] With the development of artificial intelligence technology, some fields have emerged with business opportunity recommendation solutions based on machine learning methods or deep learning methods. However, in the construction industry, since business opportunities in the construction engineering field involve multi-dimensional data, business opportunity recommendation methods in other fields are not applicable to the construction industry. Currently, most business opportunity screening in the construction engineering field is done manually, which is inefficient. Summary of the invention

[0004] In view of this, the present invention provides a method, apparatus, computer equipment and storage medium for recommending construction business opportunities to solve the problem of low efficiency in recommending construction business opportunities.

[0005] In a first aspect, the present invention provides a method for recommending construction business opportunities, comprising:

[0006] Acquire historical business opportunity data of historical engineering projects, and acquire bid data of bid objects, wherein the bid data includes entity data and personnel data of the bid objects;

[0007] Associating the historical business opportunity data with the bidding data to generate original data; the original data includes the associated historical business opportunity data and bidding data;

[0008] Preprocessing the original data and setting corresponding labels to generate sample data with labels;

[0009] Model training is performed based on the sample data to generate a business opportunity recommendation model; the business opportunity recommendation model is used to recommend corresponding business opportunity data to the target object.

[0010] In some optional implementations, associating the historical business opportunity data with the bidding data to generate original data includes:

[0011] Determine a plurality of candidate bidders related to the historical engineering project according to the bidding process data of the historical engineering project, and determine entity data of the candidate bidders;

[0012] When the entity data of the bidding object matches the entity data of one of the candidate winning bidders, the bidding data of the bidding object is associated with the historical business opportunity data of the historical engineering project to generate corresponding original data.

[0013] In some optional implementations, the obtaining of historical business opportunity data of historical engineering projects includes:

[0014] Acquire first business opportunity data released in the first phase; the first business opportunity data includes first project information of a corresponding engineering project;

[0015] Acquire second business opportunity data released in the second phase; the second business opportunity data includes second project information of the corresponding engineering project, and the second phase is later than the first phase;

[0016] determining a similarity between the second project information and the first project information;

[0017] When the similarity between the second project information and the first project information exceeds a preset similarity and is greater than the similarity between the second project information and the project information of any other business opportunity data released in the first phase, determining that the second business opportunity data is associated with the first business opportunity data;

[0018] According to the associated first business opportunity data and second business opportunity data, historical business opportunity data of the corresponding engineering project is generated.

[0019] In some optional implementations, determining the similarity between the second project information and the first project information includes:

[0020] Determining a plurality of common fields contained in the first project information and the second project information;

[0021] Vectorizing the information corresponding to each of the common fields in the first item information to generate a first feature vector; vectorizing the information corresponding to each of the common fields in the second item information to generate a second feature vector;

[0022] The similarity between the second feature vector and the first feature vector is used as the similarity between the second item information and the first item information.

[0023] In some optional implementations, preprocessing the raw data and setting corresponding labels to generate sample data with labels includes:

[0024] Filter out valid data corresponding to valid fields from the original data; the valid fields are fields determined based on correlation analysis and are related to distinguishing whether a bid is successful;

[0025] Performing feature engineering processing on the valid data to generate sample data for input into the model;

[0026] According to the bidding results of the bidding objects in the original data, corresponding labels are set for the sample data.

[0027] In some optional implementations, the process of determining the valid field includes:

[0028] At least one extended field is set for the data to be analyzed; the method of generating the data to be analyzed is the same as the method of generating the original data;

[0029] Assigning values ​​to the extended fields according to data related to historical engineering projects and / or bidding objects of the data to be analyzed, to obtain extended data to be analyzed;

[0030] Using the expanded fields in the data to be analyzed as fields to be determined, and determining the correlation between each field to be determined and whether the bid is successful;

[0031] taking at least part of the fields to be judged whose correlation exceeds a preset threshold as valid fields;

[0032] The step of filtering out valid data corresponding to valid fields from the original data includes:

[0033] In the case where some valid fields are extended fields, valid extended fields are set for the original data and assigned values ​​to obtain the expanded original data.

[0034] In some optional implementations, performing model training according to the sample data to generate a business opportunity recommendation model includes:

[0035] Divide the sample data into multiple data domains; wherein the historical business opportunity data, the entity data of the bidding object, and the personnel data of the bidding object correspond to different data domains respectively;

[0036] A model supporting input of data from multiple data domains is trained based on the sample data, and the associations between different data domains are learned to generate a business opportunity recommendation model.

[0037] In some optional embodiments, the method further comprises:

[0038] Obtain the current business opportunity data of the current project, and obtain the entity data and personnel data of the target object;

[0039] Generate current data; the current data includes the current business opportunity data, entity data and personnel data of the target object;

[0040] Preprocessing the current data, inputting the preprocessed current data into the business opportunity recommendation model, and determining the winning probability of the target object;

[0041] Determine whether to recommend the current business opportunity data to the target object according to the bid winning probability.

[0042] In a second aspect, the present invention provides a device for recommending construction business opportunities, comprising:

[0043] A data acquisition module, used to acquire historical business opportunity data of historical engineering projects and acquire bidding data of bidding objects, wherein the bidding data includes entity data and personnel data of the bidding objects;

[0044] A data association module, used to associate the historical business opportunity data with the bid data to generate original data; the original data includes the associated historical business opportunity data and bid data;

[0045] A processing module, used to pre-process the raw data, set corresponding labels, and generate sample data with labels;

[0046] The training module is used to perform model training based on the sample data to generate a business opportunity recommendation model; the business opportunity recommendation model is used to recommend corresponding business opportunity data to the target object.

[0047] In a third aspect, the present invention provides a computer device, comprising: a memory and a processor, the memory and the processor being communicatively connected to each other, the memory storing computer instructions, and the processor executing the method for recommending construction business opportunities according to the first aspect or any corresponding embodiment thereof by executing the computer instructions.

[0048] In a fourth aspect, the present invention provides a computer-readable storage medium having computer instructions stored thereon, the computer instructions being used to cause a computer to execute the method for recommending construction business opportunities according to the first aspect or any corresponding embodiment thereof.

[0049] The present invention forms multi-dimensional structured raw data by associating the historical business opportunity data of historical engineering projects with the bidding data of bidding objects. The model is subsequently trained based on the raw data, so that the model can learn the features related to the engineering projects, so that the trained business opportunity recommendation model can recommend business opportunities in the construction field, greatly improving the efficiency of business personnel in obtaining high-quality business opportunity data. In addition, the original collected data is input into the business opportunity recommendation model in an end-to-end manner, and the business opportunity recommendation results are output, and the system deployment is also relatively simple. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the related technologies, the drawings required for use in the specific embodiments or the related technical descriptions will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0051] Figure 1 is a flow chart of a method for recommending construction business opportunities according to an embodiment of the present invention;

[0052] Figure 2 is a flow chart of another construction business opportunity recommendation method according to an embodiment of the present invention;

[0053] Figure 3 is a flow chart of data governance according to an embodiment of the present invention;

[0054] Figure 4 is a schematic diagram of correlation analysis according to an embodiment of the present invention;

[0055] Figure 5 is a schematic diagram of feature engineering and sample engineering processing according to an embodiment of the present invention;

[0056] Figure 6 is a schematic diagram of a framework of a DeepFM model according to an embodiment of the present invention;

[0057] Figure 7 is a schematic diagram of a process for implementing business opportunity recommendation based on a model according to an embodiment of the present invention;

[0058] Figure 8 is a structural block diagram of a device for recommending building business opportunities according to an embodiment of the present invention;

[0059] Fig. 9 It is a schematic diagram of the hardware structure of a computer device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0060] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0061] In the business opportunity recommendation scenario, business opportunities can be recommended to users based on traditional machine learning methods; for example, through data processing and feature engineering, an XGBoost model can be built to match the scores of user and business opportunity features, and high-scoring business opportunities can be recommended to users. Some solutions use deep learning technologies such as LSTM (Long Short-Term Memory) and CNN (Convolutional Neural Networks) for recommendation.

[0062] In the construction field, business opportunities usually refer to the bidding information of construction projects. Construction companies can use this bidding information to determine whether the project is in their own interests and then decide whether to participate in the bidding. Through business opportunity recommendations, appropriate bidding information can be recommended to construction companies, which will increase the probability of winning the bid.

[0063] However, business opportunity recommendation using traditional machine learning methods such as XGBoost requires first analyzing user portrait features, and then combining the business opportunity features to uniformly input the model to obtain the business opportunity score. This method is not suitable for the construction industry. On the one hand, the "user" here is a construction company, which is not suitable for building user features using traditional user portrait analysis methods. On the other hand, business opportunity matching in the construction field involves data from multiple dimensions such as regional policies, project types, project scales, owner information, personnel requirements, enterprise qualification requirements, and project risks. Traditional machine learning methods require manual construction of cross-features between various features in multiple dimensions, which is a huge workload, and it is impossible to extract deeper implicit relationships between features, making it difficult to further improve the results.

[0064] As for the methods of recommending business opportunities using deep learning models such as LSTM and CNN, these methods need to be coordinated with other algorithm models, and the business opportunities are preliminarily sorted and then refined in steps. The overall structure is relatively complex, and results cannot be obtained directly after data input. It is also not suitable for the multi-dimensional structured business opportunity data in the construction industry.

[0065] At present, most business opportunities in the construction engineering field are screened manually, and whether the business opportunities are suitable for the enterprise is completely judged by the experience of professionals. For example, professionals manually screen suitable regions, project scales, qualification requirements, personnel requirements and other factors on relevant bidding websites to filter out business opportunity information. On the one hand, obtaining information through manual search is relatively one-sided and inefficient, which reduces the timeliness of enterprises to quickly obtain business opportunities. On the other hand, there is a certain threshold for professionals familiar with this field, and their experience is not easy to learn, which is not conducive to the rapid learning of internal personnel of the enterprise. At the same time, business opportunities in the construction engineering field involve regional policies, project types, project scales, owner information, personnel requirements, enterprise qualification requirements, project risks and other aspects and dimensions. It is difficult to screen and determine which of the massive business opportunities are suitable for their own enterprises based on personnel experience alone.

[0066] The method for recommending construction business opportunities provided by an embodiment of the present invention associates multi-dimensional data in the construction field to form multi-dimensional structured data. The multi-dimensional structured data can be used to train a business opportunity recommendation model that can recommend construction business opportunities. Subsequently, business opportunity data can be automatically recommended based on the business opportunity recommendation model, greatly improving the efficiency of business personnel in obtaining high-quality business opportunity data.

[0067] According to an embodiment of the present invention, a method embodiment for recommending construction business opportunities is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0068] In this embodiment, a method for recommending construction business opportunities is provided. Figure 1 is a flow chart of a method for recommending construction business opportunities according to an embodiment of the present invention. Figure 1 As shown, the process includes the following steps.

[0069] Step S101, obtaining historical business opportunity data of historical engineering projects, and obtaining bidding data of bidding objects, wherein the bidding data includes entity data and personnel data of the bidding objects.

[0070] In this embodiment, in order to implement model training, it is first necessary to obtain relevant information of previous engineering projects. For the convenience of description, the engineering projects that have completed bidding are referred to as historical engineering projects.

[0071] Among them, business opportunity data related to the historical engineering project, i.e., historical business opportunity data, can be obtained; the historical business opportunity data can specifically include the project name, project location, project amount, project scale, bid deadline, etc. of the historical engineering project. Specifically, the corresponding historical business opportunity data can be collected through information such as the bidding announcement of the historical engineering project issued by the authoritative organization, or the historical business opportunity data of the historical engineering project can be obtained based on a third-party database.

[0072] In addition, there are enterprises or units that have made bids before, that is, bidding objects. The entity data of bidding objects and the personnel data of internal personnel of bidding objects can also be obtained. For the convenience of description, the entity data and personnel data of bidding objects are collectively referred to as bidding data. Among them, the entity data represents the relevant data of entities such as enterprises or units. For example, the entity data may include the entity name, registered capital, business location, business scope, personnel scale, whether there are legal risks, etc. of the bidding object; similarly, the personnel data may specifically include the data of personnel in the relevant construction field. For example, the personnel data may include the name, number, and number of historical projects undertaken by senior engineers.

[0073] It can be understood that since the training model requires a large amount of data, historical business opportunity data of multiple historical engineering projects can be obtained, and bidding data of multiple bidding objects can be obtained.

[0074] Step S102, associating the historical business opportunity data with the bid data to generate original data; the original data includes the associated historical business opportunity data and bid data.

[0075] In this embodiment, since the historical business opportunity data and the bid data may come from different data sources, it is necessary to integrate the data from different data sources. Specifically, each historical business opportunity data and each bid data are associated, so that the associated historical business opportunity data and bid data are used as a set of data, that is, raw data. Specifically, the raw data can be structured data, so that the subsequent model can extract features from the raw data.

[0076] For example, based on the bid opening records of historical engineering projects (i.e., bidding information), it is possible to determine which entities have bid for the historical engineering project, and to extract the entity data of the bid entities, such as the entity name, etc.; by comparing the entity data of the bid entities with the entity data of each bidding object, it is possible to determine which bidding objects are associated with the historical business opportunity data of the historical engineering project.

[0077] It can be understood that, since there are generally multiple objects bidding on the same engineering project, the historical business opportunity data of a historical engineering project can be associated with the bid data of multiple bidding objects; and, an object can also bid on multiple engineering projects, so the bid data of a bidding object can also be associated with the historical business opportunity data of multiple historical engineering projects, depending on the actual situation.

[0078] Step S103, preprocessing the original data, and setting corresponding labels to generate sample data with labels.

[0079] In this embodiment, since the original data includes historical business opportunity data and bidding data, and both the historical business opportunity data and bidding data include multiple data, in order to enable the subsequent model to be better trained, the original data can be preprocessed to process the original data into sample data that is more suitable for model training. For example, feature engineering processing can be performed on the original data to remove outliers, etc., so as to generate the required sample data.

[0080] In addition, the sample data is also provided with a corresponding label. In this embodiment, the labels of the sample data are divided into two types, namely, winning and failing. Among them, it is possible to determine which bidder or bidders have won the bid based on the winning announcement of the historical engineering project, and label the sample data corresponding to the winning bidder with a label indicating "winning"; correspondingly, label the sample data corresponding to other bidders that have failed to win the bid with a label indicating "failing to win".

[0081] Step S104, performing model training based on the sample data to generate a business opportunity recommendation model; the business opportunity recommendation model is used to recommend corresponding business opportunity data to the target object.

[0082] After obtaining sample data with labels, the preset model can be trained based on these sample data, so that the model can learn the relationship between the features in the sample data and the corresponding labels, and finally train a model that can recommend corresponding business opportunity data to the target object, that is, the business opportunity recommendation model. The target object is the object that needs to use the business opportunity recommendation model for business opportunity recommendation in the future.

[0083] Moreover, when the target object uses the business opportunity recommendation model, there is no need to perform multiple model training steps, and end-to-end output results can be achieved through a single business opportunity recommendation model.

[0084] The method for recommending construction business opportunities provided in this embodiment associates the historical business opportunity data of historical engineering projects with the bidding data of bidding objects, thereby forming multi-dimensional structured raw data. Subsequently, model training based on the raw data can enable the model to learn features related to engineering projects, so that the trained business opportunity recommendation model can recommend business opportunities in the construction field, greatly improving the efficiency of business personnel in obtaining high-quality business opportunity data. In addition, the original collected data is input into the business opportunity recommendation model in an end-to-end manner, and the business opportunity recommendation results are output, and the system deployment is also relatively simple.

[0085] In this embodiment, a method for recommending construction business opportunities is provided. Figure 2 is a flow chart of a method for recommending construction business opportunities according to an embodiment of the present invention. Figure 2 As shown, the process includes the following steps.

[0086] Step S201, obtaining historical business opportunity data of historical engineering projects, and obtaining bidding data of bidding objects, wherein the bidding data includes entity data and personnel data of the bidding objects.

[0087] For details, please see Figure 1 Step S101 of the illustrated embodiment will not be described in detail here.

[0088] In some optional implementations, as described above, in the field of construction, relevant data is generally obtained from different data sources, such as relevant data (such as entity data, personnel data, etc.) from within the enterprise, or relevant data can be obtained through external industry data sources. In addition, when obtaining historical business opportunity data, since the bidding process of the project is divided into multiple stages, such as the bidding stage, the bidding stage, the candidate announcement stage, the winning bid announcement stage, etc., each stage also corresponds to corresponding business opportunity data; in order to generate complete and accurate historical business opportunity data, the business opportunity data of multiple stages are associated in this embodiment.

[0089] Specifically, the above step S201 of "obtaining historical business opportunity data of historical engineering projects" may include the following steps A1 to A5.

[0090] Step A1, obtaining the first business opportunity data released in the first phase; the first business opportunity data includes the first project information of the corresponding engineering project.

[0091] Step A2, obtaining the second business opportunity data released in the second phase; the second business opportunity data includes the second project information of the corresponding engineering project, and the second phase is later than the first phase.

[0092] In this embodiment, business opportunity data released in different stages can be obtained. For the convenience of description, the business opportunity data released in the first stage is called the first business opportunity data, and the business opportunity data released in the second stage is called the second business opportunity data; and the second stage is later than the first stage, that is, the first business opportunity data is released first and the second business opportunity data is released later.

[0093] For example, the first stage may be the bidding stage and the second stage may be the tendering stage, or the first stage may be the bidding stage and the second stage may be the bidding winning stage, etc.

[0094] Furthermore, the business opportunity data of each stage includes project information related to the engineering project, which may specifically include project name, project number, project amount, project requirements, name of the bidding object, name of the agency object (i.e., the name of the agency), etc. For ease of description, the project information in the first business opportunity data and the second business opportunity data are respectively referred to as the first project information and the second project information.

[0095] Step A3: determining the similarity between the second item information and the first item information.

[0096] In this embodiment, the similarity between the second project information and the first project information can be calculated, and based on the similarity, it can be determined whether the second project information and the first project information correspond to the same engineering project, thereby determining whether the first business opportunity data and the second business opportunity data can be associated.

[0097] Optionally, the above step A3 of “determining the similarity between the second project information and the first project information” may specifically include the following steps A31 to A33.

[0098] Step A31: determine a plurality of common fields contained in the first project information and the second project information.

[0099] Step A32, vectorize the information corresponding to each common field in the first project information to generate a first feature vector; vectorize the information corresponding to each common field in the second project information to generate a second feature vector.

[0100] Step A33: taking the similarity between the second feature vector and the first feature vector as the similarity between the second item information and the first item information.

[0101] In this embodiment, due to the publishing habits of different websites, the same engineering project may have different names at different stages and may not necessarily publish a unified project number, so it is sometimes impossible to fully match the project name or number; and because the business opportunity data published at different stages are slightly different, for example, more detailed business opportunity data will generally be published at the bidding stage, while at the candidate publicity stage and other stages, only part of the data may be published, such as the project amount and project requirements may not be published, resulting in the project information obtained at different stages may contain different fields. Based on this, first determine the multiple common fields contained in the first project information and the second project information, that is, the first project information and the second project information both contain content related to the common field. For example, the common fields may include the project name, the name of the bidding object, etc.

[0102] After determining the common fields, the information corresponding to each common field in each project information is vectorized to generate corresponding feature vectors, namely the first feature vector and the second feature vector. Then, the similarity between the second feature vector and the first feature vector can be calculated, and the similarity can be used as the similarity between the second project information and the first project information.

[0103] Step A4: if the similarity between the second project information and the first project information exceeds a preset similarity and is greater than the similarity between the second project information and the project information of any other business opportunity data released in the first phase, it is determined that the second business opportunity data is associated with the first business opportunity data.

[0104] Step A5: Generate historical business opportunity data of the corresponding engineering project according to the associated first business opportunity data and second business opportunity data.

[0105] In this embodiment, since there are multiple business opportunity data released in the first stage, each business opportunity data corresponds to a different engineering project; in order to accurately associate, it is necessary to calculate the similarity between the second project information in the second stage and the project information of each business opportunity data released in the first stage. If the similarity between the second project information and the first project information in the first stage is the highest, that is, it is greater than the similarity between the second project information and the project information of any other business opportunity data released in the first stage, and the similarity also exceeds the preset similarity (for example, 90%, 95%, etc.), it can be determined that the second project information and the first project information correspond to the same engineering project, so the second business opportunity data and the first business opportunity data are associated, and then based on the first business opportunity data and the second business opportunity data, the historical business opportunity data of the corresponding engineering project can be generated.

[0106] It can be understood that since the bidding process generally has more than two stages, the business opportunity data in any two stages can be associated based on the above steps A1 to A5, so that the business opportunity data in multiple stages of a certain project can be associated together to form complete and accurate historical business opportunity data.

[0107] In this embodiment, the project information in the business opportunity data of different stages is used to determine the similarity, so that the multi-stage business opportunity data belonging to the same engineering project can be associated to form complete and accurate historical business opportunity data, and realize data serial association. In addition, the similarity is calculated using the information corresponding to the common fields in the project information, which can ensure the accuracy of the serial association.

[0108] Optionally, the collected data is generally stored and calculated in a data warehouse. However, due to the large number of data sources, the data standards, formats, scopes, and encodings are diverse. Based on data governance measures, data from different data sources can be integrated and standardized for use in subsequent steps. The data after governance and calculation forms a unified associated data table stored in the data warehouse. Figure 3 A schematic diagram of a data governance process is shown. It can be understood that in the subsequent reasoning process, the newly collected current business opportunity data also needs to be governed by the same data.

[0109] like Figure 3 As shown, historical business opportunity data, bidding data, etc. can be cleaned, and non-standard data in each data source can be cleaned and eliminated; for example, test data or invalid data can be cleaned uniformly, or they can be cleaned separately according to the characteristics of data from different sources; for example, if there is a separate field mark or related content has the meaning of "test", the data is eliminated as test data, or non-target area, non-time range, non-construction field data is eliminated, and invalid data such as policies, business opportunities, qualifications, etc. are filtered. At the same time, data with the same content, such as enterprise information, personnel information, business opportunity information, etc., can be deduplicated according to the name or related ID.

[0110] In addition, unified standards can be made for relevant codes, region names, numerical precision, date formats, monetary units, etc. to achieve standardized data processing and ensure that the data used in subsequent steps has unified measurement, format or precision, which facilitates the subsequent data analysis and feature engineering process.

[0111] like Figure 3 As shown, after data cleaning, the collected data can be divided into business opportunity data, regional data, entity data, personnel data, etc., and the required original data can be generated by associating the related data in subsequent steps.

[0112] Step S202, associating the historical business opportunity data with the bid data to generate original data; the original data includes the associated historical business opportunity data and bid data.

[0113] For details, please see Figure 1 Step S102 of the illustrated embodiment will not be described in detail here.

[0114] In some optional implementations, the above step S202 of "associating historical business opportunity data with bid data to generate original data" may include the following steps B1 to B2.

[0115] Step B1, determining a plurality of candidate bidders related to the historical engineering project based on the bidding process data of the historical engineering project, and determining entity data of the candidate bidders.

[0116] Step B2: when the entity data of the bidding object matches the entity data of one of the candidate bidders, the bidding data of the bidding object is associated with the historical business opportunity data of the historical engineering project to generate corresponding original data.

[0117] In this embodiment, when associating historical business opportunity data and bidding data, it can be based on whether the entity data therein matches; if the entity data corresponding to the two match, it can be considered that the two can be associated. Among them, for a historical engineering project, since there are multiple (even more than ten) bidding objects, but generally there is only one winning object, therefore, for this historical engineering project, a positive sample indicating a winning bid and a large number of negative samples indicating a losing bid can be generated, resulting in an imbalance in the ratio of positive and negative samples, affecting the subsequent model training effect.

[0118] In bidding scenarios in the construction field, the results of the winning candidates are generally announced, that is, multiple winning candidates are announced, data association is performed based on these winning candidates, and positive and negative sample data are generated.

[0119] Specifically, for a certain historical engineering project, the bidding process data of each stage can be obtained, including the data of the bid candidate announcement stage, based on which the bid candidate objects can be extracted, and the number of the bid candidate objects is generally multiple. In addition, the entity data of the bid candidate objects can be obtained from the bidding process data or other means, and the entity data can specifically include the name, business location, business scope, etc. of the bid candidate objects.

[0120] It can be determined which bidding object's entity data matches the entity data of the candidate winning object. For example, it can be determined whether the two match based on whether the names of the objects are consistent or by calculating similarity. If the entity data of a bidding object matches the entity data of one of the candidate winning objects, it can be determined that the bidding object is a candidate winning object of the historical engineering project. Therefore, the bidding data of the bidding object can be associated with the historical business opportunity data of the historical engineering project to generate a corresponding original data.

[0121] When setting labels subsequently, the sample data corresponding to the final winning object among the winning candidate objects may be set with a winning label, while the sample data corresponding to other winning candidate objects may be set with a failing label.

[0122] For example, if three candidate bidders for a certain project are announced during the candidate bid announcement stage, then for the project, one positive sample data and two negative sample data can be generated in the end, which greatly balances the ratio of positive and negative samples and avoids data imbalance.

[0123] Step S203, preprocessing the original data, and setting corresponding labels to generate sample data with labels.

[0124] For details, please see Figure 1 Step S103 of the illustrated embodiment will not be described in detail here.

[0125] In some optional implementations, the above step S203 of “preprocessing the original data, setting corresponding labels, and generating sample data with labels” may include the following steps C1 to C3.

[0126] Step C1, filtering out valid data corresponding to valid fields from the original data; valid fields are fields determined based on correlation analysis and are related to distinguishing whether a bid is successful.

[0127] In this embodiment, since the original data contains data of a large number of fields, the data of some fields may not be related to the bid, so the data of these fields can be removed, and only the fields related to distinguishing whether the bid is successful, that is, the valid fields, are retained. For example, if the original data contains fields such as "Contact Phone Number", this field is unlikely to affect the bid result, so the field can be deleted, that is, "Contact Phone Number" is not a valid field.

[0128] Among them, the fields that belong to the valid fields can be determined in advance through correlation analysis. Optionally, the process of determining the valid fields can include steps D1 to D4.

[0129] Step D1, setting at least one extended field for the data to be analyzed; the method of generating the data to be analyzed is the same as the method of generating the original data.

[0130] Step D2, assigning values ​​to the extended fields according to the data related to the historical engineering projects and / or bidding objects of the data to be analyzed, to obtain the extended data to be analyzed.

[0131] In this embodiment, it is necessary to screen out appropriate valid fields based on certain data, and the data is referred to as data to be analyzed. The method of generating the data to be analyzed is the same as the method of generating the original data; for example, the data to be analyzed can be generated based on the above steps S201 to S202; for example, a part of the original data can also be directly used as the data to be analyzed.

[0132] For the data to be analyzed, in addition to the existing fields, one or more extended fields can be determined based on human experience or other methods to facilitate subsequent model training, and values ​​can be assigned to them; for example, extended fields can be designed according to business needs or statistics of a certain dimension can be used as extended fields. For example, fields for large, medium and small scales of projects can be added based on the bidding amount and the company's own situation; statistical information under different historical time windows, such as the number of bids submitted by a company within three months, the number of large projects released in a region within one month, etc.

[0133] Step D3, taking the expanded fields in the data to be analyzed as fields to be determined, and determining the correlation between each field to be determined and whether the bid is successful.

[0134] Step D4: taking at least part of the fields to be determined whose correlation exceeds a preset threshold as valid fields.

[0135] In this embodiment, the expanded data to be analyzed includes previously existing fields and expanded fields. These fields are all used as fields to be judged, and the correlation between each field to be judged and whether the bid is successful is determined separately, so that some fields to be judged with higher correlation can be used as valid fields.

[0136] The essence of business opportunity recommendation for construction companies is to determine whether a project can be awarded. Therefore, it is necessary to analyze the possible correlation between the winning or losing results of a certain company's bid and the corresponding fields of each dimension. In addition to expert experience, correlation analysis, hypothesis testing and simple models in statistical analysis can also be used to determine the importance of features, thereby screening out valid fields.

[0137] Figure 4 A schematic diagram of correlation analysis of the fields to be judged is shown. Figure 4As shown, for a field to be judged, a correlation analysis can be performed. According to the field to be judged and the result of winning / not winning the bid, the correlation can be determined by the correlation coefficient (such as Pearson coefficient or Kendall coefficient, etc.) for subsequent judgment. It is also possible to use the hypothesis test method, according to the characteristics of different fields to be judged, to count the mean variance and other statistics of the field to be judged, select a suitable hypothesis test method (such as chi-square test or T test), calculate the P value of the significant difference with the winning result, and judge the significance level of the field to be judged on the winning result based on this. In addition, model feature screening can also be performed: specifically, a certain batch of data samples can be constructed, with the features of all fields to be judged and the labels of winning or not winning the bid. The XGBoost model can be trained according to the binary classification problem, and the feature ranking can be output according to the feature importance through the model structure characteristics.

[0138] After the above processing, feature screening can be performed to remove the fields that have no correlation or weak correlation in each step, and the remaining fields can be used as valid fields. After the valid fields are screened out, these valid fields can be directly used as feature inputs for the original data in the subsequent process without the need for correlation analysis.

[0139] Furthermore, the above-mentioned step C1 of "screening out valid data corresponding to valid fields from original data" may specifically include step C11.

[0140] Step C11, when some valid fields are extended fields, valid extended fields are set for the original data and assigned values ​​to obtain the expanded original data.

[0141] Among them, if some valid fields are extended fields set in step D1, when filtering out valid data corresponding to the valid fields from the original data, it is necessary to first perform field expansion on the original data, that is, set a valid extended field (an extended field belonging to the valid field) for it and assign a value to it, so as to obtain the expanded original data.

[0142] Step C2, perform feature engineering on the valid data to generate sample data for input into the model.

[0143] Step C3, setting corresponding labels for the sample data according to the bidding results of the bidding objects in the original data.

[0144] In this embodiment, feature engineering is performed on valid data to generate sample data suitable for input into the model. In addition, sample engineering is performed based on the bidding results of the bidding objects in the original data, so as to set corresponding labels for the sample data, and finally generate sample data with labels.

[0145] Figure 5A schematic diagram of feature engineering and sample engineering processing for valid data is shown. Figure 5 As shown in the figure, when performing feature engineering on valid data, the null value rate of feature fields can be counted, null values ​​can be filled, outliers can be removed or replaced, and categorical data can be encoded with one-hot or hash encoding; numerical data can be normalized or standardized, and bucketed into categorical features using quantiles or business experience, and then encoded with reference to the categorical data encoding method, such as one-hot encoding. Through feature engineering processing, sample data models can be trained faster and better.

[0146] In addition, for each sample data, a corresponding label can be marked according to the bidding results of the bidding objects in the original data; for example, the label of the winning object is 1, plus its corresponding feature data is a positive sample, and the label of the unsuccessful object is 0, plus its corresponding feature data is a negative sample. The positive and negative samples can be balanced based on the method of step B1 to step B2. In addition, after the sample data is sorted out, the training samples and test samples can be divided into 80% / 20% or 70% / 30%, and the model can be trained based on the training samples, and the trained model can be tested based on the test samples.

[0147] Step S204, performing model training based on the sample data to generate a business opportunity recommendation model; the business opportunity recommendation model is used to recommend corresponding business opportunity data to the target object.

[0148] Specifically, the above step S204 "performing model training according to sample data to generate a business opportunity recommendation model" includes steps S2041 to S2042.

[0149] Step S2041, dividing the sample data into multiple data domains; wherein the historical business opportunity data, the entity data of the bidding object, and the personnel data of the bidding object correspond to different data domains respectively.

[0150] Step S2042: training a model that supports input of data from multiple data domains based on sample data, learning the associations between different data domains, and generating a business opportunity recommendation model.

[0151] In this embodiment, unlike the way that the input data is set as a certain sequence in models such as LSTM, the multi-dimensional sample data is divided into multiple data domains, and each dimension corresponds to a data domain; specifically, as mentioned above, the historical business opportunity data, the entity data of the bidding object, and the personnel data of the bidding object correspond to different data domains. Among them, when generating the original data, the corresponding regional data can also be obtained based on the regional location of the historical engineering project. The regional data can specifically include economic information, future planning information, GDP, etc. of the area where the historical engineering project is located; in addition, when generating the original data, the entity data related to the bidding object can also be added, and accordingly, the regional data and the entity data of the bidding object can also be used as different data domains.

[0152] In addition, this embodiment uses a model that supports the input of data from multiple data domains, such as a DeepFM (Deep Factorization Machine) model, and then uses data from multiple data domains in the sample data to train the model, so that the model can learn the associations between different data domains, thereby enabling the trained business opportunity recommendation model to utilize the associations between multiple data domains and more accurately output the probability of winning the bid.

[0153] Figure 6 A schematic diagram of the DeepFM model is shown. The DeepFM model includes an embedding layer, an FM part, a Deep part, and an output layer. The embedding layer is an embedding layer. The sample data obtained after one-hot or hash coding is sparse, and the feature input dimension is large. After the embedding layer, it can be converted into a dense feature vector, which reduces the feature dimension and enhances the feature meaning expression, which is convenient for subsequent calculations. Different feature domains correspond to different data domains, which are used to input data in the corresponding data domains. The FM part can calculate the first-order relationship and second-order cross relationship of the feature, learn the weights of the direct and second-order cross relationship of the feature, and intuitively reflect the direct impact of the feature on the business opportunity matching results in each dimension. The Deep part uses a deep neural network to construct multiple hidden layers, which can learn the deep-level relationship weights of the features, can go deep into the deep correlation between the features in each dimension, and extract the deep implicit cross features to calculate the impact on the business opportunity matching. The calculation results of the FM part and the Deep part are input to the output layer, and finally the binary classification output results can be calculated to determine whether the bid is successful, which can meet the effective calculation of the features on the business opportunity results in multiple dimensions in the construction field.

[0154] Step S205, obtaining the current business opportunity data of the current project, and obtaining the entity data and personnel data of the target object.

[0155] Step S206, generating current data; the current data includes current business opportunity data, entity data of the target object, and personnel data.

[0156] Step S207, pre-processing the current data, inputting the pre-processed current data into the business opportunity recommendation model, and determining the probability of winning the bid for the target object.

[0157] Step S208: Determine whether to recommend the current business opportunity data to the target object according to the bid winning probability.

[0158] In this embodiment, after the business opportunity recommendation model is trained, the business opportunity recommendation model can be used to recommend appropriate business opportunity data to the target object. Specifically, when business opportunity recommendation is required, the current engineering project that can be bid is determined, that is, the current engineering project, and accordingly, the business opportunity data of the current engineering project, that is, the current business opportunity data, can be obtained.

[0159] In addition, the entity data and personnel data of the target object can be obtained based on the internal database of the target object. For example, if you need to recommend business opportunities to enterprise A, you can use enterprise A as the target object and obtain the entity data (enterprise name, registered capital, staff size, etc.) and personnel data (such as the number of senior engineers, etc.) of enterprise A.

[0160] Afterwards, similar to the process of generating sample data, current data can be generated by combining current business opportunity data, entity data of the target object, and personnel data. The current data is then preprocessed to generate data suitable for input into a business opportunity recommendation model.

[0161] The process of obtaining the current business opportunity data and preprocessing the current data is similar to the process of obtaining the historical business opportunity data and preprocessing the original data, and will not be repeated here. It can be understood that when generating the current data, no association is required, that is, the current business opportunity data, the entity data of the target object, and the personnel data are directly integrated to generate the current data; and in the process of preprocessing the current data, the data related to the valid fields can be directly extracted, without the need to re-determine the valid fields, and no sample engineering processing is required afterwards.

[0162] In this embodiment, when the model is trained, the labels used are winning or not winning, so the business opportunity recommendation model obtained by training can output the corresponding winning probability; specifically, if the pre-processed current data is input into the business opportunity recommendation model, the predicted winning probability can be obtained if the target object bids for the current engineering project corresponding to the current business opportunity data. For example, the greater the winning probability, the more likely the target object is to win the bid, so the current business opportunity data with a higher winning probability can be recommended to the target object. For example, the top 10 or top 20 current business opportunity data with a higher winning probability can be recommended for the target object to choose.

[0163] Figure 7 Figure 2 shows a flow chart of implementing business opportunity recommendation based on the model. Figure 7 As shown in Figure 1, internal enterprise data, external industry data, etc. can be collected. In the training process, after data governance (for example, data cleaning), feature engineering, sample engineering ( Figure 7 The feature engineering, sample engineering and other processes are not shown in the figure) to generate appropriate sample data. When the model is trained for the first time, if no appropriate valid fields are determined, data analysis is required to extract valid fields. The DeepFM model is constructed, the generated sample data is input into the model, the effect is evaluated according to the indicators, the model parameters are adjusted, and a model with better effect is obtained through multiple iterations to obtain a business opportunity recommendation model.

[0164] In the inference process, the collected data also needs to be processed through data governance and feature engineering, and then input into the trained business opportunity recommendation model. After inference calculation, the score of the corresponding business opportunity data, that is, the probability of winning the bid, can be obtained. The scores of each business opportunity data can be ranked and recommended to relevant personnel in the business system. In addition, business opportunity recommendation verification can also be performed to determine whether the recommended business opportunity data is appropriate, thereby verifying the accuracy of the business opportunity recommendation model.

[0165] The method for recommending construction business opportunities provided in this embodiment associates the historical business opportunity data of historical engineering projects with the bidding data of bidding objects, thereby forming multi-dimensional structured original data. The subsequent model training based on the original data can enable the model to learn the features related to the engineering projects, so that the trained business opportunity recommendation model can recommend business opportunities in the construction field, greatly improving the efficiency of business personnel in obtaining high-quality business opportunity data. In addition, the original collected data is input into the business opportunity recommendation model in an end-to-end manner, and then the business opportunity recommendation results are output, and the system deployment is also relatively simple. Using DeepFM models and other models that support multiple data domains for training and evaluation, the weights of the relationship between features can be learned to match construction companies and business opportunities, so that a better business opportunity recommendation method can be obtained, which can meet the effective calculation of business opportunity results by features in multiple dimensions in the construction field, and is conducive to learning to obtain better business opportunity matching effects.

[0166] In this embodiment, a device for recommending construction business opportunities is also provided, which is used to implement the above embodiments and preferred implementations, and the descriptions that have been made will not be repeated. As used below, the term "module" can be a combination of software and / or hardware that implements a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceivable.

[0167] This embodiment provides a device for recommending construction business opportunities, such as Figure 8 As shown, including:

[0168] The data acquisition module 801 is used to acquire historical business opportunity data of historical engineering projects and acquire bidding data of bidding objects, wherein the bidding data includes entity data and personnel data of the bidding objects;

[0169] The data association module 802 is used to associate the historical business opportunity data with the bid data to generate original data; the original data includes the associated historical business opportunity data and bid data;

[0170] The processing module 803 is used to pre-process the raw data and set corresponding labels to generate sample data with labels;

[0171] The training module 804 is used to perform model training according to the sample data to generate a business opportunity recommendation model; the business opportunity recommendation model is used to recommend corresponding business opportunity data to the target object.

[0172] In some optional implementations, the data association module 802 associates the historical business opportunity data with the bid data to generate original data, including:

[0173] Determine a plurality of candidate bidders related to the historical engineering project according to the bidding process data of the historical engineering project, and determine entity data of the candidate bidders;

[0174] When the entity data of the bidding object matches the entity data of one of the candidate winning bidders, the bidding data of the bidding object is associated with the historical business opportunity data of the historical engineering project to generate corresponding original data.

[0175] In some optional implementations, the data acquisition module 801 acquires historical business opportunity data of historical engineering projects, including:

[0176] Acquire first business opportunity data released in the first phase; the first business opportunity data includes first project information of a corresponding engineering project;

[0177] Acquire second business opportunity data released in the second phase; the second business opportunity data includes second project information of the corresponding engineering project, and the second phase is later than the first phase;

[0178] determining a similarity between the second project information and the first project information;

[0179] When the similarity between the second project information and the first project information exceeds a preset similarity and is greater than the similarity between the second project information and the project information of any other business opportunity data released in the first phase, determining that the second business opportunity data is associated with the first business opportunity data;

[0180] According to the associated first business opportunity data and second business opportunity data, historical business opportunity data of the corresponding engineering project is generated.

[0181] In some optional implementations, the data acquisition module 801 determines the similarity between the second project information and the first project information, including:

[0182] Determining a plurality of common fields contained in the first project information and the second project information;

[0183] Vectorizing the information corresponding to each of the common fields in the first item information to generate a first feature vector; vectorizing the information corresponding to each of the common fields in the second item information to generate a second feature vector;

[0184] The similarity between the second feature vector and the first feature vector is used as the similarity between the second item information and the first item information.

[0185] In some optional implementations, the processing module 803 pre-processes the raw data and sets corresponding labels to generate sample data with labels, including:

[0186] Filter out valid data corresponding to valid fields from the original data; the valid fields are fields determined based on correlation analysis and are related to distinguishing whether a bid is successful;

[0187] Performing feature engineering processing on the valid data to generate sample data for input into the model;

[0188] According to the bidding results of the bidding objects in the original data, corresponding labels are set for the sample data.

[0189] In some optional implementations, the process of determining the valid field includes:

[0190] At least one extended field is set for the data to be analyzed; the method of generating the data to be analyzed is the same as the method of generating the original data;

[0191] Assigning values ​​to the extended fields according to data related to historical engineering projects and / or bidding objects of the data to be analyzed, to obtain extended data to be analyzed;

[0192] Using the expanded fields in the data to be analyzed as fields to be determined, and determining the correlation between each field to be determined and whether the bid is successful;

[0193] taking at least part of the fields to be judged whose correlation exceeds a preset threshold as valid fields;

[0194] The processing module 803 filters out valid data corresponding to valid fields from the original data, including:

[0195] In the case where some valid fields are extended fields, valid extended fields are set for the original data and assigned values ​​to obtain the expanded original data.

[0196] In some optional implementations, the training module 804 performs model training according to the sample data to generate a business opportunity recommendation model, including:

[0197] Divide the sample data into multiple data domains; wherein the historical business opportunity data, the entity data of the bidding object, and the personnel data of the bidding object correspond to different data domains respectively;

[0198] A model supporting input of data from multiple data domains is trained based on the sample data, and the associations between different data domains are learned to generate a business opportunity recommendation model.

[0199] In some optional implementations, the device further includes a recommendation module, configured to:

[0200] Obtain the current business opportunity data of the current project, and obtain the entity data and personnel data of the target object;

[0201] Generate current data; the current data includes the current business opportunity data, entity data and personnel data of the target object;

[0202] Preprocessing the current data, inputting the preprocessed current data into the business opportunity recommendation model, and determining the winning probability of the target object;

[0203] Determine whether to recommend the current business opportunity data to the target object according to the bid winning probability.

[0204] The further functional description of each of the above modules and units is the same as that of the above corresponding embodiments and will not be repeated here.

[0205] The device for recommending construction business opportunities in this embodiment is presented in the form of a functional unit, where the unit refers to an ASIC (Application Specific Integrated Circuit) circuit, including a processor and memory that executes one or more software or fixed programs, and / or other devices that can provide the above functions.

[0206] The embodiment of the present invention also provides a computer device having the above Figure 8 The building opportunities shown are recommended installations.

[0207] See also Fig. 9 , Fig. 9 is a schematic diagram of the structure of a computer device provided by an optional embodiment of the present invention, such as Fig. 9 As shown, the computer device includes: one or more processors 10, a memory 20, and interfaces for connecting various components, including high-speed interfaces and low-speed interfaces. Various components are connected to each other using different buses for communication, and can be installed on a common mainboard or installed in other ways as needed. The processor can process instructions executed in the computer device, including instructions stored in or on the memory to display the graphical information of the GUI on an external input / output device (such as a display device coupled to the interface). In some optional embodiments, if necessary, multiple processors and / or multiple buses can be used together with multiple memories. Similarly, multiple computer devices can be connected, and each device provides some necessary operations (for example, as a server array, a group of blade servers, or a multi-processor system). Fig. 9 A processor 10 is taken as an example.

[0208] The processor 10 may be a central processing unit, a network processor or a combination thereof. The processor 10 may further include a hardware chip. The hardware chip may be a dedicated integrated circuit, a programmable logic device or a combination thereof. The programmable logic device may be a complex programmable logic device, a field programmable gate array, a general purpose array logic or any combination thereof.

[0209] The memory 20 stores instructions executable by at least one processor 10, so that the at least one processor 10 executes the method shown in the above embodiment.

[0210] The memory 20 may include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function; the data storage area may store data created according to the use of the computer device, etc. In addition, the memory 20 may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some optional embodiments, the memory 20 may optionally include a memory remotely arranged relative to the processor 10, and these remote memories may be connected to the computer device via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0211] The memory 20 may include a volatile memory, such as a random access memory; the memory may also include a non-volatile memory, such as a flash memory, a hard disk or a solid state drive; the memory 20 may also include a combination of the above types of memory.

[0212] The computer device also includes an input device 30 and an output device 40. The processor 10, the memory 20, the input device 30 and the output device 40 may be connected via a bus or other means. Fig. 9 The example of connecting through bus is taken in the following.

[0213] The input device 30 can receive input digital or character information, and generate key signal input related to the user settings and function control of the computer device, such as a touch screen, a keypad, a mouse, a track pad, a touch pad, an indicator bar, one or more mouse buttons, a trackball, a joystick, etc. The output device 40 may include a display device, an auxiliary lighting device (e.g., an LED) and a tactile feedback device (e.g., a vibration motor), etc. The above-mentioned display device includes but is not limited to a liquid crystal display, a light emitting diode, a display and a plasma display. In some optional embodiments, the display device can be a touch screen.

[0214] The embodiment of the present invention also provides a computer-readable storage medium. The method according to the embodiment of the present invention can be implemented in hardware, firmware, or can be implemented as a computer code that can be recorded in a storage medium, or can be implemented as a computer code that is originally stored in a remote storage medium or a non-temporary machine-readable storage medium and will be stored in a local storage medium through a network download, so that the method described herein can be stored in such software processing on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. Among them, the storage medium can be a magnetic disk, an optical disk, a read-only storage memory, a random access memory, a flash memory, a hard disk or a solid-state hard disk, etc.; further, the storage medium can also include a combination of the above types of memories. It can be understood that a computer, a processor, a microprocessor controller, or programmable hardware includes a storage component that can store or receive software or computer code. When the software or computer code is accessed and executed by a computer, a processor, or hardware, the method shown in the above embodiment is implemented.

[0215] A part of the present invention may be applied as a computer program product, such as a computer program instruction, which, when executed by a computer, can call or provide the method and / or technical solution according to the present invention through the operation of the computer. Those skilled in the art should understand that the existence of the computer program instruction in a computer-readable medium includes, but is not limited to, a source file, an executable file, an installation package file, etc., and accordingly, the way in which the computer program instruction is executed by the computer includes, but is not limited to: the computer directly executes the instruction, or the computer compiles the instruction and then executes the corresponding compiled program, or the computer reads and executes the instruction, or the computer reads and installs the instruction and then executes the corresponding installed program. Here, the computer-readable medium may be any available computer-readable storage medium or communication medium accessible to the computer.

[0216] Although the embodiments of the present invention have been described in conjunction with the accompanying drawings, those skilled in the art may make various modifications and variations without departing from the spirit and scope of the present invention, and such modifications and variations are all within the scope defined by the appended claims.

Claims

1. A method for recommending construction business opportunities, characterized in that: The method comprises: Acquire historical business opportunity data of historical engineering projects, and acquire bid data of bid objects, wherein the bid data includes entity data and personnel data of the bid objects; Associating the historical business opportunity data with the bidding data to generate original data; the original data includes the associated historical business opportunity data and bidding data; Preprocessing the original data and setting corresponding labels to generate sample data with labels; Model training is performed based on the sample data to generate a business opportunity recommendation model; the business opportunity recommendation model is used to recommend corresponding business opportunity data to the target object.

2. The method according to claim 1, characterized in that The step of associating the historical business opportunity data with the bidding data to generate original data includes: Determine a plurality of candidate bidders related to the historical engineering project according to the bidding process data of the historical engineering project, and determine entity data of the candidate bidders; When the entity data of the bidding object matches the entity data of one of the candidate winning bidders, the bidding data of the bidding object is associated with the historical business opportunity data of the historical engineering project to generate corresponding original data.

3. The method according to claim 1, characterized in that The obtaining of historical business opportunity data of historical engineering projects includes: Acquire first business opportunity data released in the first phase; the first business opportunity data includes first project information of a corresponding engineering project; Acquire second business opportunity data released in the second phase; the second business opportunity data includes second project information of the corresponding engineering project, and the second phase is later than the first phase; determining a similarity between the second project information and the first project information; When the similarity between the second project information and the first project information exceeds a preset similarity and is greater than the similarity between the second project information and the project information of any other business opportunity data released in the first phase, determining that the second business opportunity data is associated with the first business opportunity data; According to the associated first business opportunity data and second business opportunity data, historical business opportunity data of the corresponding engineering project is generated.

4. The method according to claim 3, characterized in that The determining the similarity between the second project information and the first project information includes: Determining a plurality of common fields contained in the first project information and the second project information; Vectorizing the information corresponding to each of the common fields in the first item information to generate a first feature vector; vectorizing the information corresponding to each of the common fields in the second item information to generate a second feature vector; The similarity between the second feature vector and the first feature vector is used as the similarity between the second item information and the first item information.

5. The method according to claim 1, characterized in that The preprocessing of the original data and setting corresponding labels to generate sample data with labels includes: Filter out valid data corresponding to valid fields from the original data; the valid fields are fields determined based on correlation analysis and are related to distinguishing whether a bid is successful; Performing feature engineering processing on the valid data to generate sample data for input into the model; According to the bidding results of the bidding objects in the original data, corresponding labels are set for the sample data.

6. The method according to claim 5, characterized in that The process of determining the valid fields includes: At least one extended field is set for the data to be analyzed; the method of generating the data to be analyzed is the same as the method of generating the original data; Assigning values ​​to the extended fields according to data related to historical engineering projects and / or bidding objects of the data to be analyzed, to obtain extended data to be analyzed; Using the expanded fields in the data to be analyzed as fields to be determined, and determining the correlation between each field to be determined and whether the bid is successful; taking at least part of the fields to be judged whose correlation exceeds a preset threshold as valid fields; The step of filtering out valid data corresponding to valid fields from the original data includes: In the case where some valid fields are extended fields, valid extended fields are set for the original data and assigned values ​​to obtain the expanded original data.

7. The method according to claim 1, characterized in that The performing model training according to the sample data to generate a business opportunity recommendation model includes: Divide the sample data into multiple data domains; wherein the historical business opportunity data, the entity data of the bidding object, and the personnel data of the bidding object correspond to different data domains respectively; A model supporting input of data from multiple data domains is trained based on the sample data, and the associations between different data domains are learned to generate a business opportunity recommendation model.

8. The method according to claim 1, characterized in that The method further comprises: Obtain the current business opportunity data of the current project, and obtain the entity data and personnel data of the target object; Generate current data; the current data includes the current business opportunity data, entity data and personnel data of the target object; Preprocessing the current data, inputting the preprocessed current data into the business opportunity recommendation model, and determining the winning probability of the target object; Determine whether to recommend the current business opportunity data to the target object according to the bid winning probability.

9. A device for recommending construction business opportunities, characterized in that: The device comprises: A data acquisition module, used to acquire historical business opportunity data of historical engineering projects and acquire bidding data of bidding objects, wherein the bidding data includes entity data and personnel data of the bidding objects; A data association module, used to associate the historical business opportunity data with the bid data to generate original data; the original data includes the associated historical business opportunity data and bid data; A processing module, used to pre-process the raw data, set corresponding labels, and generate sample data with labels; The training module is used to perform model training based on the sample data to generate a business opportunity recommendation model; the business opportunity recommendation model is used to recommend corresponding business opportunity data to the target object.

10. A computer device, characterized in that: include: A memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the method for recommending construction business opportunities according to any one of claims 1 to 8 by executing the computer instructions.

11. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a computer to execute the method for recommending construction business opportunities according to any one of claims 1 to 8.

Images