Competitive product information analysis method and device, electronic equipment and storage medium

Abstract

The invention discloses a competitive product information analysis method and device, electronic equipment and a storage medium. The method comprises the steps that positioning information of a to-be-analyzed target item is acquired; obtaining multi-dimensional target feature data of the target item and the plurality of candidate items from an item database based on the positioning information of the target item in real time; wherein the project database comprises multi-dimensional target feature data, obtained and updated from a plurality of data sources, of each project through a distributed data acquisition cluster deployed in real time; inputting the target feature data of each candidate item and the target item into a double-tower deep learning model, performing feature processing on the target feature data of each candidate item and the target feature data of the target item, and matching the processed feature data to obtain the overall similarity of the plurality of candidate items and the target item; and based on the overall similarity between each candidate item and the target item, screening out the current competitive item of the plurality of target items.

Description

Technical Field

[0001] This application relates to the field of big data analysis technology, and in particular to a method and apparatus for competitor information analysis, electronic equipment, and storage medium. Background Technology

[0002] In real estate and other commercial projects, for a particular project, due to differences in project positioning, not all projects will affect that project; that is, not all projects are competitors of that project. Therefore, it is necessary to analyze the information of each project in order to identify the competitors of a project.

[0003] Currently, staff primarily collect information on projects within the same defined region as the project being analyzed through offline surveys and online searches. Then, an expert system analyzes the information for each project, relying on expert experience to identify competitors based on the analysis results. This competitor information is then provided to the user to offer decision-making support.

[0004] However, the current method of manually collecting data not only fails to provide timely analysis based on dynamic market changes, leading to delayed results, but also limits analysis to a small number of projects within a limited area each time, resulting in incomplete and inaccurate results. Furthermore, the data from various dimensions is not effectively integrated and analyzed, and the reliance on manual selection introduces strong subjectivity and inconsistent standards, leading to inaccurate analysis results. Summary of the Invention

[0005] In view of the shortcomings of the prior art, this application provides a competitor information analysis method, apparatus, electronic device, and storage medium to solve the problem that the prior art cannot efficiently and accurately analyze competitor information.

[0006] To achieve the above objectives, this application provides the following technical solution:

[0007] The first aspect of this application provides a method for competitor information analysis, including:

[0008] Obtain the location information of the target project to be analyzed;

[0009] Based on the location information of the target project, multi-dimensional target feature data of the target project and multiple candidate projects are obtained from the project database in real time; wherein, the project database includes multi-dimensional target feature data of each project obtained and updated in real time from multiple data sources through a deployed distributed data acquisition cluster;

[0010] The target feature data of each candidate project and the target project are input into the dual-tower deep learning model. Feature processing is performed on the target feature data of each candidate project and the target project respectively. The processed feature data are then matched to obtain the overall similarity between the candidate projects and the target project.

[0011] Based on the overall similarity between each candidate project and the target project, multiple current competitor projects of the target project are selected.

[0012] Optionally, the above-mentioned competitor information analysis method also includes:

[0013] In real time, through distributed data acquisition clusters deployed on various data sources, structured and unstructured information of each project published in each data source is collected to obtain target feature data of each project.

[0014] According to the rules in the data cleaning rule base, abnormal data in the target feature data of each project is cleaned.

[0015] The project database is updated using the cleaned target feature data of each project.

[0016] Optionally, in the above-described competitor information analysis method, after obtaining multi-dimensional target feature data of the target project and multiple candidate projects from the project database in real time based on the target project's location information, the method further includes:

[0017] The geographic features in the target feature data of the target project and each of the candidate projects are enhanced to obtain the enhanced geographic features of the target project and each of the candidate projects.

[0018] The enhanced geographic features of the target project and each of the candidate projects are respectively added to their target feature data.

[0019] Optionally, in the above-described competitor information analysis method, the step of inputting the target feature data of each candidate project and the target project into a dual-tower deep learning model, performing feature processing on the target feature data of each candidate project and the target project respectively, and matching the processed feature data to obtain the overall similarity between multiple candidate projects and the target project includes:

[0020] The target feature data of the target project and the target feature data of each candidate project are respectively input into the two tower model branches in the dual-tower deep learning model;

[0021] The target feature data is processed by the two branches of the tower model to obtain the feature vector of the target item and the feature vector of each candidate item.

[0022] Based on the feature vector of the target project and the feature vector of each candidate project output by the dual-tower deep learning model, multiple target candidate projects most similar to the target project are retrieved by the approximate nearest neighbor algorithm.

[0023] The feature vector of the target project and each feature data in the feature vector of each target candidate project are matched to obtain the overall similarity between each target candidate project and the target project.

[0024] Optionally, in the above-described competitor information analysis method, the step of matching the feature vector of the target project with each feature data in the feature vector of each of the target candidate projects to obtain the overall similarity between each of the target candidate projects and the target project includes:

[0025] The similarity of each feature data in the feature vector of the target project and the feature vector of each target candidate project is matched to obtain the similarity of each feature data. The similarity of each feature data is then weighted and summed to obtain the overall similarity between each target candidate project and the target project.

[0026] Optionally, in the above-described competitor information analysis method, the step of filtering out multiple current competitor projects of the target project based on the overall similarity between each candidate project and the target project includes:

[0027] Each candidate project whose overall similarity to the target project and its corresponding confidence level are both greater than the corresponding preset threshold is identified as the current competitor project of the target project.

[0028] Optionally, in the above-described competitor information analysis method, after filtering out multiple current competitor projects of the target project based on the overall similarity between each candidate project and the target project, the method further includes:

[0029] Output a list of names of each current competitor project, the target feature data, the overall similarity with the target project, and its confidence level;

[0030] Display the similarity and contribution of each current competitor project to the target project in each dimension of feature data;

[0031] Based on the similarity of the feature data of each current competitor project and the target project in each dimension, a comparison radar chart of each current competitor project is generated and displayed;

[0032] Based on the similarity of the feature data of each current competitor project and the target project in each dimension, competitive strategy suggestions are matched and displayed;

[0033] Anomaly monitoring is performed on the feature data of each of the current competing products across multiple specified dimensions.

[0034] A second aspect of this application provides a competitor information analysis device, comprising:

[0035] The location information acquisition unit is used to acquire the location information of the target project to be analyzed.

[0036] The feature data acquisition unit is used to acquire multi-dimensional target feature data of the target project and multiple candidate projects from the project database in real time based on the location information of the target project; wherein, the project database includes multi-dimensional target feature data of each project acquired and updated in real time from multiple data sources through a deployed distributed data acquisition cluster;

[0037] The matching unit is used to input the target feature data of each candidate item and the target item into the dual-tower deep learning model, perform feature processing on the target feature data of each candidate item and the target feature data of the target item respectively, and match the processed feature data to obtain the overall similarity between multiple candidate items and the target item.

[0038] The competitor screening unit is used to screen out multiple current competitor projects of the target project based on the overall similarity between each candidate project and the target project.

[0039] Optionally, the aforementioned competitor information analysis device further includes:

[0040] The data acquisition unit is used to collect structured and unstructured information of each project published in each data source in real time through a distributed data acquisition cluster deployed in each data source, and obtain the target feature data of each project.

[0041] The data cleaning unit is used to clean abnormal data in the target feature data of each project according to the rules in the data cleaning rule base;

[0042] The data update unit is used to update the project database using the cleaned target feature data of each of the projects.

[0043] Optionally, the aforementioned competitor information analysis device further includes:

[0044] A feature enhancement unit is used to enhance the geographic features in the target feature data of the target project and each of the candidate projects respectively, so as to obtain the enhanced geographic features of the target project and each of the candidate projects;

[0045] The feature enhancement unit is used to add enhanced geographic features of the target project and each of the candidate projects to their target feature data, respectively.

[0046] Optionally, in the above-described competitor information analysis device, the matching unit includes:

[0047] The input unit is used to input the target feature data of the target project and the target feature data of each candidate project into the two tower model branches in the dual-tower deep learning model, respectively.

[0048] The feature processing unit is used to perform feature processing on the input target feature data through the two branches of the tower model respectively, to obtain the processed feature vector of the target item and the feature vector of each candidate item;

[0049] The project screening unit is used to retrieve multiple target candidate projects that are most similar to the target project based on the feature vector of the target project and the feature vector of each candidate project output by the dual-tower deep learning model, using an approximate nearest neighbor algorithm.

[0050] The similarity calculation unit is used to match the feature vector of the target item with each feature data in the feature vector of each target candidate item to obtain the overall similarity between each target candidate item and the target item.

[0051] Optionally, in the above-mentioned competitor information analysis device, the similarity calculation unit includes:

[0052] The weighted calculation unit is used to perform similarity matching on each feature data in the feature vector of the target project and the feature vector of each target candidate project to obtain the similarity of each feature data, and to perform weighted summation on the similarity of each feature data to obtain the overall similarity between each target candidate project and the target project.

[0053] Optionally, in the above-mentioned competitor information analysis device, the competitor screening unit includes:

[0054] The competitor screening subunit is used to identify each candidate project whose overall similarity and corresponding confidence level with the target project are both greater than a preset threshold as the current competitor project of the target project.

[0055] Optionally, the aforementioned competitor information analysis device further includes:

[0056] The information output unit is used to output a list of names of each current competitor project, the target feature data, the overall similarity with the target project and its confidence level;

[0057] The first display unit is used to display the similarity and contribution of the feature data of each current competitor project and the target project in each dimension;

[0058] The second display unit is used to generate and display a comparison radar chart of each current competitor project based on the similarity of feature data of each current competitor project and the target project in each dimension;

[0059] The suggestion unit is used to match competitive strategy suggestion information with the similarity of feature data of each current competitor project and the target project in each dimension and then display it.

[0060] The monitoring unit is used to monitor for anomalies in the feature data of each of the current competing products across multiple specified dimensions.

[0061] A third aspect of this application provides an electronic device, comprising:

[0062] Memory and processor;

[0063] The memory is used to store programs;

[0064] The processor is used to execute the program, which, when executed, is specifically used to implement the competitor information analysis method as described in any of the above.

[0065] A fourth aspect of this application provides a computer storage medium for storing a computer program, which, when executed by a processor, is used to implement the competitor information analysis method as described in any of the preceding claims.

[0066] This application provides a competitor information analysis method. It utilizes a deployed distributed data acquisition cluster to acquire and update multi-dimensional target feature data of various projects from multiple data sources in real time, storing this data in a project database. This automatically obtains comprehensive data, improves data acquisition efficiency, and allows for the analysis of a large number of projects, yielding accurate results. When competitor analysis is required, the method acquires the location information of the target project. Based on the target project's location information, it acquires multi-dimensional target feature data of the target project and multiple candidate projects from the project database in real time. Then, the target feature data of each candidate project and the target project are input into a dual-tower deep learning model. Feature processing is performed on the target feature data of each candidate project and the target project, respectively. The processed feature data is then matched to obtain the overall similarity between multiple candidate projects and the target project. Finally, based on the overall similarity between each candidate project and the target project, the current competitor projects of multiple target projects are selected. Thus, the dual-tower deep learning model efficiently analyzes the competitors of the target project, eliminating reliance on manual analysis and providing accurate results. Therefore, it implements a method for efficiently and accurately analyzing competitors. Attached Figure Description

[0067] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of this application. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0068] Figure 1 A flowchart illustrating a competitor information analysis method provided in this application embodiment;

[0069] Figure 2 A flowchart illustrating a method for collecting project data provided in this application embodiment;

[0070] Figure 3 A flowchart for project matching provided in this application embodiment;

[0071] Figure 4 A flowchart illustrating a method for displaying and monitoring information on competitor projects, provided in an embodiment of this application;

[0072] Figure 5 A schematic diagram of the architecture of a competitor information analysis device provided in this application embodiment;

[0073] Figure 6 This is a schematic diagram of the architecture of an electronic device provided in an embodiment of this application. Detailed Implementation

[0074] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0075] In this application, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0076] This application provides a method for competitor information analysis, such as... Figure 1 As shown, it includes the following steps:

[0077] S101. Obtain the location information of the target project to be analyzed.

[0078] Optionally, the user can input the location information of the target project to be analyzed or select the map coordinates of the target project and input the location information of the target project in order to determine the target project to be analyzed and the candidate projects to be analyzed.

[0079] S102. Based on the location information of the target project, obtain multi-dimensional target feature data of the target project and multiple candidate projects from the project database in real time.

[0080] The target feature data refers to the specified feature data. The project database includes multi-dimensional target feature data for each project, which is acquired and updated in real time from multiple data sources through a deployed distributed data acquisition cluster. Therefore, in this embodiment, a distributed data acquisition cluster is deployed in multiple data sources, thereby automatically collecting multi-dimensional information about each project from each data source in real time. For example, for real estate projects, data such as price, plot ratio, unit type, and area can be collected.

[0081] It should be noted that because projects are constantly being added or removed, and the information of existing projects is constantly changing, project data is continuously collected and updated in the project database to obtain accurate results adaptively. Furthermore, based on the target project's location information, multi-dimensional target feature data of the target project and multiple candidate projects are retrieved from the project database in real time to obtain the latest project information and analyze the acquired data to identify competitors.

[0082] Specifically, the system identifies target projects in real time based on their location information and retrieves multi-dimensional target feature data from the project database. Optionally, since projects located far away typically do not affect the target project, projects within a certain range of the target project can be identified as candidate projects based on its location information, and multi-dimensional target feature data for each candidate project can be retrieved from the project database. Because this application automatically collects and analyzes project information without requiring manual analysis, a wider range of projects can be selected as candidate projects.

[0083] Optionally, in another embodiment of this application, a method for collecting project data is further provided, such as... Figure 2 As shown, it includes:

[0084] S201. In real time, through distributed data acquisition clusters deployed on various data sources, the structured and unstructured information of each project published in each data source is collected to obtain the target feature data of each project.

[0085] The structured information of the project consists of the clearly defined features from the data source, such as price, plot ratio, unit type, and area. Unstructured data requires further analysis, such as NLP feature extraction from the project description text.

[0086] S202. Clean the abnormal data in the target feature data of each project according to the rules in the data cleaning rule base.

[0087] It should be noted that since there may be anomalous data in the acquired feature data, which may affect the accuracy of the analysis results, it is necessary to clean the anomalous data in the target feature data of each project according to the rules in the data cleaning rule base.

[0088] S203. Update the project database using the target feature data of each project after cleaning.

[0089] Specifically, for projects whose target feature data already exists in the project database, the data of those projects in the project database is updated using the target feature data of the current project. For projects whose target feature data exists in the current project database but not in the project database, the target feature data of the current project is added to the project database.

[0090] Optionally, in order to make the features in the target feature data of the project more comprehensive and thus improve the accuracy of the analysis results, in another embodiment of this application, after performing step S102, the following is further included:

[0091] The geographic features in the target feature data of the target project and each candidate project are enhanced to obtain the enhanced geographic features of the target project and each candidate project, and the enhanced geographic features of the target project and each candidate project are added to their target feature data respectively.

[0092] It should be noted that the target feature data includes the project's geographical features, such as its latitude and longitude. However, while geographical features are crucial for a project, the information conveyed by a single project location is limited. Therefore, it is necessary to add more geographical features to extract more relevant surrounding geographical information, such as distances to subway stations, commercial areas, schools, and neighborhood environmental statistics. This data is then added to the project's target feature data, enriching its dimensions.

[0093] S103. Input the target feature data of each candidate project and the target project into the dual-tower deep learning model, perform feature processing on the target feature data of each candidate project and the target project respectively, and match the processed feature data to obtain the overall similarity between multiple candidate projects and the target project.

[0094] It should be noted that, in order to efficiently match candidate and target items, this embodiment employs a dual-tower deep learning model. Two branches of the established tower model process the target feature data of both the target and candidate items, respectively. Based on the processed target feature data, the two are then matched to obtain the overall similarity between multiple candidate items and the target item. Optionally, the target item can be matched with all candidate items to calculate the overall similarity between all candidate items and the target item. Alternatively, only the overall similarity between the most similar candidate item and the target item can be calculated.

[0095] Optionally, in another embodiment of this application, one specific implementation of step S103 is as follows: Figure 3 As shown, it includes:

[0096] S301. Input the target feature data of the target project and the target feature data of each candidate project into the two tower model branches in the dual-tower deep learning model.

[0097] It should be noted that two tower model branch networks are designed in the dual-tower deep learning model, which are used to process the target item and the candidate item respectively. Therefore, the target feature data of the target item is input into one tower model branch, and the target feature data of the candidate item is input into the other tower model branch.

[0098] S302. The input target feature data is processed through two tower model branches to obtain the feature vectors of the target item and the feature vectors of each candidate item.

[0099] In order to perform project matching using the input target feature data, it is necessary to perform feature processing on the input target feature data through the tower model branch, thereby processing the target feature data into feature vectors.

[0100] S303. Based on the feature vector of the target project and the feature vector of each candidate project output by the dual-tower deep learning model, the approximate nearest neighbor algorithm is used to retrieve multiple target candidate projects that are most similar to the target project.

[0101] It should be noted that, since this embodiment allows for the analysis of a large number of items, resulting in a large number of candidate items, an approximate nearest neighbor algorithm is first used to retrieve the multiple candidate items most similar to the target item, which are then selected as the target candidate items to improve analysis efficiency.

[0102] S304. Match the feature vector of the target project with the feature vectors of each target candidate project to obtain the overall similarity between each target candidate project and the target project.

[0103] Specifically, a matching network can be used to match each target candidate item, each feature data in the feature vector of each target item and the target candidate, and integrate the matching results of each feature data to obtain the overall similarity between the target subsequent items and the target item.

[0104] Optionally, in another embodiment of this application, a specific implementation of step S304 includes:

[0105] The matching network in the dual-tower deep learning model performs similarity matching on the feature vector of the target project and the feature vector of each target candidate project to obtain the similarity of each feature data. The similarity of each feature data is then weighted and summed to obtain the overall similarity between each target candidate project and the target project.

[0106] Since the importance of feature data varies across different dimensions, similarity matching is performed on the data in each dimension to obtain the similarity of feature data in each dimension. Then, according to the preset weights of feature data in each dimension, the similarity of feature data in each dimension is weighted and summed to obtain the overall similarity between each target candidate item and the target item.

[0107] S104. Based on the overall similarity between each candidate project and the target project, select the current competing projects of multiple target projects.

[0108] Specifically, each candidate project can be sorted according to its overall similarity to the target project, and then multiple candidate projects with an overall similarity greater than a preset threshold can be selected as the current competitors of the target project.

[0109] Optionally, in another embodiment of this application, one specific implementation of step S104 includes:

[0110] Candidate projects whose overall similarity to the target project and their corresponding confidence scores are both greater than the corresponding preset thresholds are identified as the current competitors of the target project.

[0111] Since the overall similarity is obtained through model analysis, to ensure its accuracy, it is necessary not only to select candidate projects with an overall similarity greater than a preset similarity threshold, but also to ensure that the confidence level corresponding to their overall similarity is greater than a preset confidence threshold, thus identifying the current competitors of the target project. For example, candidate projects with a similarity > 85% and a confidence level > 90% are identified as the current competitors of the target project.

[0112] Optionally, in another embodiment of this application, a method for displaying and monitoring information on competing products is further included, such as... Figure 4 As shown, it includes:

[0113] S401. Output a list of names of each current competitor project, target feature data, overall similarity with the target project, and its confidence level.

[0114] In order to enable users to understand the attribute information and related analysis data of each current competitor project, it is necessary not only to output the list of names and target feature data of each current competitor project, but also to output its overall similarity and confidence level with the target project, so that users can understand the various similar competitor projects.

[0115] S402. Display the similarity and contribution of the feature data of each current competitor project and the target project in each dimension.

[0116] In order to display more accurate data and make accurate decisions, in addition to displaying the overall data, it is also necessary to display the similarity and sharing of features in each dimension.

[0117] S403. Based on the similarity of the feature data of each current competitor project and the target project in each dimension, generate and display a comparison radar chart of each current competitor project.

[0118] To allow users to more intuitively understand the similarity of competitor projects and target projects in various dimensions of feature data, a comparative radar chart of each current competitor project is generated and displayed by utilizing the similarity of feature data of the current competitor project and target project in each dimension.

[0119] S404. Match competitive strategy suggestions based on the similarity of feature data of each current competitor project and the target project in each dimension, and display them.

[0120] In order to better assist users in making decisions, the decisions that need to be made when correcting feature data are pre-configured. Therefore, competitive strategy suggestions can be matched and displayed based on the similarity of feature data of each current competitor project and the target project in each dimension.

[0121] S405. Perform anomaly monitoring on the feature data of multiple specified dimensions of each current competitor project.

[0122] Because competitor data is constantly changing, to ensure users are notified promptly of significant changes and can make timely adjustments, we also monitor for anomalies in multiple specified dimensions of feature data for each competitor. An alert is triggered when feature data exceeds a specified threshold. For example, monitoring competitor pricing and triggering an alert when the rate of price change exceeds a threshold.

[0123] This application provides a competitor information analysis method. It utilizes a deployed distributed data acquisition cluster to acquire and update multi-dimensional target feature data of various projects from multiple data sources in real time, storing it in a project database. This automatically acquires comprehensive data, improves data acquisition efficiency, and allows for the analysis of a large number of projects, yielding accurate results. When competitor analysis is required, the method acquires the location information of the target project. Based on the target project's location information, it acquires multi-dimensional target feature data of the target project and multiple candidate projects from the project database in real time. Then, the target feature data of each candidate project and the target project are input into a dual-tower deep learning model. Feature processing is performed on the target feature data of each candidate project and the target project, and the processed feature data is matched to obtain the overall similarity between multiple candidate projects and the target project. Finally, based on the overall similarity between each candidate project and the target project, the current competitor projects of multiple target projects are selected. Thus, the dual-tower deep learning model efficiently analyzes the competitors of the target project, eliminating reliance on manual analysis and providing accurate results. Therefore, it implements a method for efficiently and accurately analyzing competitors.

[0124] Another embodiment of this application provides a competitor information analysis device, such as... Figure 5 As shown, it includes:

[0125] The location information acquisition unit 501 is used to acquire the location information of the target project to be analyzed.

[0126] The feature data acquisition unit 502 is used to acquire multi-dimensional target feature data of the target project and multiple candidate projects from the project database in real time based on the location information of the target project. The project database includes multi-dimensional target feature data of each project acquired and updated in real time from multiple data sources through a deployed distributed data acquisition cluster.

[0127] The matching unit 503 is used to input the target feature data of each candidate project and the target project into the dual-tower deep learning model, perform feature processing on the target feature data of each candidate project and the target feature data of the target project respectively, and match the processed feature data to obtain the overall similarity between multiple candidate projects and the target project.

[0128] The competitor screening unit 504 is used to filter out the current competitor projects of multiple target projects based on the overall similarity between each candidate project and the target project.

[0129] Optionally, in another embodiment of the competitor information analysis device provided in this application, the device further includes:

[0130] The data acquisition unit is used to collect structured and unstructured information of each project published in each data source in real time through a distributed data acquisition cluster deployed in each data source, so as to obtain the target feature data of each project.

[0131] The data cleaning unit is used to clean abnormal data in the target feature data of each project according to the rules in the data cleaning rule base.

[0132] The data update unit is used to update the project database using the cleaned target feature data of each project.

[0133] Optionally, in another embodiment of the competitor information analysis device provided in this application, the device further includes:

[0134] The feature enhancement unit is used to enhance the geographic features in the target feature data of the target project and each candidate project respectively, so as to obtain the enhanced geographic features of the target project and each candidate project.

[0135] The feature enhancement unit is used to add enhanced geographic features of the target project and each candidate project to their target feature data, respectively.

[0136] Optionally, in another embodiment of the competitor information analysis device provided in this application, the matching unit includes:

[0137] The input unit is used to input the target feature data of the target project and the target feature data of each candidate project into the two tower model branches in the dual-tower deep learning model, respectively.

[0138] The feature processing unit is used to process the input target feature data through two tower model branches to obtain the feature vectors of the target item and the feature vectors of each candidate item.

[0139] The project selection unit is used to retrieve multiple target candidate projects that are most similar to the target project based on the feature vector of the target project and the feature vector of each candidate project output by the dual-tower deep learning model, using the approximate nearest neighbor algorithm.

[0140] The similarity calculation unit is used to match the feature vector of the target item with the feature vector of each target candidate item to obtain the overall similarity between each target candidate item and the target item.

[0141] Optionally, in another embodiment of the competitor information analysis apparatus provided in this application, the similarity calculation unit includes:

[0142] The weighted calculation unit is used to perform similarity matching on each feature data in the feature vector of the target project and the feature vector of each target candidate project, to obtain the similarity of each feature data, and to perform weighted summation on the similarity of each feature data to obtain the overall similarity between each target candidate project and the target project.

[0143] Optionally, in another embodiment of the competitor information analysis device provided in this application, the competitor screening unit includes:

[0144] The competitor screening subunit is used to identify each candidate project whose overall similarity to the target project and its corresponding confidence level are both greater than the corresponding preset threshold as the current competitor project of the target project.

[0145] Optionally, in another embodiment of the competitor information analysis device provided in this application, the device further includes:

[0146] The information output unit is used to output a list of names of each current competitor project, target feature data, overall similarity with the target project, and its confidence level.

[0147] The first display unit is used to display the similarity and contribution of each current competitor project and the target project in each dimension of feature data.

[0148] The second display unit is used to generate and display a comparison radar chart of each current competitor project based on the similarity of feature data of each current competitor project and the target project in each dimension.

[0149] The suggestion unit is used to match competitive strategy suggestions with the target project based on the similarity of feature data in each dimension and display them.

[0150] The monitoring unit is used to monitor for anomalies in the feature data of various current competing products across multiple specified dimensions.

[0151] It should be noted that the specific working process of each unit provided in the above embodiments of this application can be referred to the implementation process of the corresponding steps in the above method embodiments, and will not be repeated here.

[0152] Another embodiment of this application provides an electronic device, such as... Figure 6 As shown, it includes:

[0153] Memory 601 and processor 602.

[0154] The memory 601 is used to store the program.

[0155] The processor 602 is used to execute the program stored in the memory 601. When the program is executed, it is specifically used to implement the competitor information analysis method provided in any of the above embodiments.

[0156] Another embodiment of this application provides a computer storage medium for storing a computer program, which, when executed by a processor, is used to implement the competitor information analysis method provided in any of the above embodiments.

[0157] Computer storage media, including both permanent and non-permanent, removable and non-removable media, can store information using any method or technology. Information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0158] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0159] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily 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 this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method of analyzing competitive information, characterized by, The method comprises the following steps: obtaining positioning information of a target item to be analyzed; obtaining multi-dimensional target feature data of the target item and a plurality of candidate items from a project database in real time based on the positioning information of the target item, wherein the project database comprises multi-dimensional target feature data of each item updated in real time by a distributed data acquisition cluster deployed at a plurality of data sources; inputting the target feature data of each of the candidate items and the target item into a double-tower deep learning model, respectively processing the target feature data of each of the candidate items and the target item, and matching the processed feature data to obtain overall similarity of the plurality of candidate items and the target item; based on the overall similarity of each of the candidate items and the target item, screening out a plurality of current competitive product items of the target item.

2. The method of claim 1, wherein, Further comprising: collecting structured information and unstructured information of each item published in each data source by a distributed data acquisition cluster deployed at each data source to obtain target feature data of each item; cleaning abnormal data in the target feature data of each item according to rules in a data cleaning rule library; updating the project database using the cleaned target feature data of each item.

3. The method of claim 1, wherein, After the step of obtaining multi-dimensional target feature data of the target item and a plurality of candidate items from a project database in real time based on the positioning information of the target item, further comprising: respectively enhancing geographical features in the target feature data of the target item and each of the candidate items to obtain enhanced geographical features of the target item and each of the candidate items; adding the enhanced geographical features of the target item and each of the candidate items to the target feature data thereof, respectively.

4. The method of claim 1, wherein, The step of inputting the target feature data of each of the candidate items and the target item into a double-tower deep learning model, respectively processing the target feature data of each of the candidate items and the target item, and matching the processed feature data to obtain overall similarity of the plurality of candidate items and the target item, comprises: inputting the target feature data of the target item and the target feature data of each of the candidate items into two tower model branches in the double-tower deep learning model, respectively; processing the input target feature data through the two tower model branches to obtain processed feature vectors of the target item and each of the candidate items; based on the feature vectors of the target item and each of the candidate items output by the double-tower deep learning model, retrieving a plurality of target candidate items most similar to the target item by an approximate nearest neighbor algorithm; matching each feature data in the feature vector of the target item and the feature vector of each of the target candidate items to obtain overall similarity of each of the target candidate items and the target item.

5. The method of claim 4, wherein, The matching of each feature data in the feature vector of the target item and the feature vector of each target candidate item obtains the overall similarity of each target candidate item to the target item, and includes: Respectively, the similarity matching of each feature data in the feature vector of the target item and the feature vector of each target candidate item obtains the similarity of each feature data, and the weighted sum of the similarity of each feature data obtains the overall similarity of each target candidate item to the target item.

6. The method of claim 1, wherein, The filtering of the current competitive product items of the target item based on the overall similarity of each candidate item to the target item includes: Each candidate item with an overall similarity to the target item greater than a corresponding preset threshold and a corresponding confidence is determined as a current competitive product item of the target item.

7. The method of claim 1, wherein, After the filtering of the current competitive product items of the target item based on the overall similarity of each candidate item to the target item, it further includes: Outputting the name list of each current competitive product item, the target feature data, the overall similarity to the target item, and the confidence thereof; Displaying the similarity and contribution of each current competitive product item to the feature data of each dimension of the target item; Generating a comparison radar chart of each current competitive product item according to the similarity of each current competitive product item to the feature data of each dimension of the target item and displaying the comparison radar chart; Matching a competitive strategy suggestion information according to the similarity of each current competitive product item to the feature data of each dimension of the target item and displaying the competitive strategy suggestion information; Abnormal monitoring of the feature data of multiple specified dimensions of each current competitive product item.

8. A rival product information analysis device characterized by comprising: It includes: A positioning information acquisition unit configured to acquire positioning information of a target item to be analyzed; A feature data acquisition unit configured to acquire, based on the positioning information of the target item, multi-dimensional target feature data of the target item and multiple candidate items from a project database in real time; wherein the project database includes multi-dimensional target feature data of each item acquired and updated from multiple data sources by a deployed distributed data acquisition cluster in real time; A matching unit configured to input the target feature data of each candidate item and the target item into a double-tower deep learning model, perform feature processing on the target feature data of each candidate item and the target item, and match the processed feature data to obtain the overall similarity of the multiple candidate items to the target item; A competitive product screening unit configured to filter, based on the overall similarity of each candidate item to the target item, multiple current competitive product items of the target item.

9. An electronic device, comprising: It includes: A memory and a processor; The memory is configured to store a program; The processor is configured to execute the program, and when the program is executed, it is specifically configured to implement the competitive product information analysis method of any one of claims 1 to 7.

10. A computer storage medium, characterized in that A computer program for storing a computer program which, when executed by a processor, implements the competitive product information analysis method according to any one of claims 1 to 7.

Images