WO2023040155A1 - 基于预设标签的策略生成方法、装置及存储介质 - Google Patents

基于预设标签的策略生成方法、装置及存储介质 Download PDF

Info

Publication number
WO2023040155A1
WO2023040155A1 PCT/CN2022/072364 CN2022072364W WO2023040155A1 WO 2023040155 A1 WO2023040155 A1 WO 2023040155A1 CN 2022072364 W CN2022072364 W CN 2022072364W WO 2023040155 A1 WO2023040155 A1 WO 2023040155A1
Authority
WO
WIPO (PCT)
Prior art keywords
product
push
information
attribute
evaluation factor
Prior art date
Application number
PCT/CN2022/072364
Other languages
English (en)
French (fr)
Inventor
林浩然
王磊
王媛
Original Assignee
平安科技（深圳）有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 平安科技（深圳）有限公司 filed Critical 平安科技（深圳）有限公司
Publication of WO2023040155A1 publication Critical patent/WO2023040155A1/zh

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
    • G06Q40/08—Insurance
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N5/00—Computing arrangements using knowledge-based models
    • G06N5/02—Knowledge representation; Symbolic representation
    • G06N5/022—Knowledge engineering; Knowledge acquisition
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q30/00—Commerce
    • G06Q30/06—Buying, selling or leasing transactions
    • G06Q30/0601—Electronic shopping [e-shopping]
    • G06Q30/0631—Item recommendations

Definitions

  • the present application relates to the field of artificial intelligence, in particular to a strategy generation method, device and storage medium based on preset tags.
  • the generation and implementation of quantitative investment strategies is a process that requires a lot of experiments, which involves factors mining, factor testing, factor combination, strategy experiment and strategy monitoring, etc.
  • the source of factors is extremely rich, including traditional financial indicators, price volume indicators, technical indicators and other factors of different frequencies, as well as indicators mined and processed from information such as texts (research reports, public opinion reviews) , so the factor processing process is complicated, and the strategy generation process also utilizes expert experience, machine learning, deep learning and other methods.
  • Many problems, such as time-varying and market style switching lead to too many factors to be considered, resulting in a decrease in screening efficiency and problems in the generation of investment strategies that are not objective and accurate enough.
  • the purpose of this application is to provide a strategy generation method, device, and storage medium based on preset labels, aiming to solve the problem that in the prior art, too many factors need to be considered in the screening of strategies, resulting in a decrease in screening efficiency and the generation of investment strategies is not objective enough.
  • the problem of insufficient accuracy is to provide a strategy generation method, device, and storage medium based on preset labels, aiming to solve the problem that in the prior art, too many factors need to be considered in the screening of strategies, resulting in a decrease in screening efficiency and the generation of investment strategies is not objective enough. The problem of insufficient accuracy.
  • the present application provides a method for generating a strategy based on preset tags, including: obtaining a preset tag input by a user, the preset tag including the target product's own attribute tag and an external risk tag; the first push product, and input the first push product into the first push pool; obtain the second push product matching the external risk label, and input the second push product into the second push pool; The union of the first push pool and the second push pool is used as the product to be pushed; the product to be pushed is input into a pre-trained prediction model to obtain the expected result information of each product to be pushed; The expected result information and the product to be pushed are pushed to the user.
  • the present application also provides an investment strategy generation device based on preset labels, including: a preset label acquisition module, used to acquire preset labels input by the user, the preset labels include the target product's own attribute labels and external Risk label; the first acquisition module is used to acquire the first push product that matches the attribute label of its own, and enters the first push product into the first push pool; the second acquisition module is used to acquire the product that is compatible with the external The second push product that the risk tag matches, and input the second push product into the second push pool; the push module is used to use the union of the first push pool and the second push pool as the product to be pushed;
  • the expected result information acquisition module is used to input the pre-trained prediction model of the products to be pushed to obtain the expected result information of each of the products to be pushed; the information push module is used to input the expected result information and the The product to be pushed is pushed to the user.
  • the present application also provides a storage medium, in which a computer program is stored, and when it is run on a computer, the computer executes the method for generating a policy based on a preset label described in the above embodiment, including: acquiring user-input Preset tags, the preset tags include the target product's own attribute tags and external risk tags; obtain the first push product that matches the self attribute tag, and input the first push product into the first push pool; Acquiring a second push product that matches the external risk label, and inputting the second push product into a second push pool; using the union of the first push pool and the second push pool as the product to be pushed; Inputting the products to be pushed into a pre-trained prediction model to obtain expected result information of each of the products to be pushed; and pushing the expected result information and the products to be pushed to the user.
  • the present application also provides a computer device containing instructions.
  • the computer device executes the preset label-based policy generation method described in the above embodiments through its internal processor, including: Obtain the preset label input by the user, which includes the target product's own attribute label and external risk label; obtain the first push product that matches the self attribute label, and input the first push product into the second A push pool; obtain a second push product that matches the external risk label, and input the second push product into the second push pool; use the union of the first push pool and the second push pool as The product to be pushed; inputting the product to be pushed into a pre-trained prediction model to obtain expected result information of each product to be pushed; pushing the expected result information and the product to be pushed to the user.
  • Beneficial effects of the present application by obtaining the preset tags input by the user, splitting the preset tags into self-attribute tags and external risk tags, and obtaining products to be pushed that match the self-attribute tags and external risk tags , and push the product to be pushed to the user after the income forecast is realized.
  • the preset label input by the user can be automatically split into its own attribute label and external risk label to achieve a more accurate generation strategy and speed up It improves the screening efficiency and solves the problem that too many factors need to be considered in the generation of existing strategies in the prior art, resulting in the decrease of screening efficiency and the problem that the generation results of strategies are not objective enough and the accuracy is not high enough, and the speed and accuracy of strategy generation are improved. sex and objectivity.
  • FIG. 1 is a schematic flow diagram of a strategy generation method based on a preset label of the present application
  • FIG. 2 is a schematic structural diagram of a policy generation device based on preset tags of the present application
  • FIG. 3 is a structural block diagram of an embodiment of a storage medium of the present application.
  • Fig. 4 is the structural block diagram of an embodiment of the computer equipment of the present application.
  • the label names in the figure are: 1-preset label acquisition module, 2-label splitting module, 3-first acquisition module, 4-second acquisition module, 5-push module, 6-expected result information acquisition module, 7- Information push module, 100-storage medium, 200-computer program, 300-computer equipment.
  • AI Artificial Intelligence
  • digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results.
  • the present application provides a policy generation method based on preset labels, including:
  • the preset tag includes the target product's own attribute tag and an external risk tag
  • the self-attribute tag is a tag associated with the target product's own attribute information
  • the The external risk label is a label associated with the target product identity information
  • the application is applied in the field of investment
  • the target product may be a stock
  • the self-attribute label includes a label associated with its own attribute information
  • the self-attribute information includes its own financial scale information, self-identity information, etc.
  • self-financial scale information includes stock market value, price-earnings ratio, total share capital, dividend rate, etc.
  • self-identity information includes stock name, stock number, place of issue, company name, etc.
  • Self-attribute tags associated with self-attribute information include high income, high rate of return, steady return, high break rate, etc.;
  • the self-attribute tags are included in a preset self-attribute tag library, and each of the self-attribute tags is matched with a corresponding first evaluation factor.
  • the first evaluation factor of the "high-income" self-attribute tag is If the return on net assets and gross profit rate of the target product mentioned above, then only when the return on net assets and gross profit rate of the matched target product are higher than the preset value, can the attribute information of the target product and "high income" be judged Self-attribute tag matching;
  • the external risk tags are included in the preset external risk tag library, and the external risk tags include tags associated with non-financial information and identity information such as external market prices and news, such as non-financial information in public opinion information and research report information. , related information of identity information, news event reports and other related tags, which can specifically include: Hong Kong stock listing, shareholder divestment, bankruptcy risk, etc.
  • Each external risk tag is matched with a corresponding second evaluation factor.
  • the corresponding second evaluation factor is the keyword “bankruptcy” and “executed”. Keywords, when words such as “bankruptcy” and “executed” appear in the matching research reports and news of the target product, the external risk information of the target product can be determined to be related to the "emergence" keyword. Bankruptcy" self-attribute tag matching;
  • the types of the self attribute label and the external risk label, the first evaluation factor of the self attribute label and the second evaluation factor of the external risk label can be set by the user, and this application does not make restrictions;
  • the server may be an independent server, or provide cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content distribution network (content Delivery Network, CDN), and cloud servers for basic cloud computing services such as big data and artificial intelligence platforms.
  • cloud services cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content distribution network (content Delivery Network, CDN), and cloud servers for basic cloud computing services such as big data and artificial intelligence platforms.
  • the server obtains the target product that matches the first evaluation factor (return on net assets, gross profit rate) in the self-attribute tag of "high income", which is specifically expressed as that the server obtains the return on net assets rate and gross profit rate are higher than the preset value, as the first push product, and the target products that are successfully matched are collected and input into the first push pool;
  • the first evaluation factor return on net assets, gross profit rate
  • the server judges that the self-attribute label of "high income” input by the user matches the first evaluation factor as "return on net assets, gross profit margin", and then the server obtains the return on net assets 5 stocks whose profit rate and gross profit rate are higher than the preset value, and integrate 5 stocks into the first push pool;
  • the external risk information includes research report information and news information
  • the second evaluation factor obtained by the server for the "bankruptcy risk” external risk label is the keyword "bankruptcy” and the keyword “executed”, and the keyword “bankruptcy” exists in the research report information and news information obtained by the server And the corresponding stocks of the "executed” keyword, and gather the successfully matched stocks to generate the second push pool. It can be understood that the information contained in the external risk information can be set by the user himself, and this application is not correct it limits;
  • the second evaluation factor of the external risk label of the "bankruptcy risk” obtained by the server is the keyword of "bankruptcy” and the keyword of "executed”, and the server obtains the research report information And there are the 7 stocks with keywords of "bankruptcy” and "executed” in the news information, and integrate the 7 stocks into the second push pool.
  • the server merges the stocks contained in the first push pool and the second push pool to generate products to be pushed;
  • the server uses 5 stocks contained in the first push pool and 12 stocks in total of 7 stocks in the second push pool as products to be pushed;
  • the server uses a preset revenue forecasting model to obtain the expected result information of each stock in the product to be pushed.
  • the modeling method of the preset revenue forecasting model It can be any one of xgb modeling or lightgbm modeling, which is not limited in this application;
  • the server uses the lightgbm modeling set by the staff to calculate the expected result information of the 12 stocks in the product to be pushed;
  • the server pushes the product to be pushed and the expected result information corresponding to each product to be pushed to the user, In this way, investment strategies are generated and pushed to corresponding users.
  • the server pushes the 12 stocks in the product to be pushed and the expected result information of the 12 stocks to the mobile phone of the user.
  • the technical effect of the present application is: by obtaining the preset tags input by the user, splitting the preset tags into self-attribute tags and external risk tags, and obtaining the to-be-pushed tags matching the self-attribute tags and external risk tags Products, and the method of predicting the income of the products to be pushed and then pushing them to the user achieves a more accurate generation of investment strategies by automatically splitting the preset tags entered by the user into their own attribute tags and external risk tags
  • the screening efficiency is accelerated, which solves the problem that too many factors need to be considered in the generation of existing investment strategies in the prior art, resulting in a decrease in screening efficiency and the problem that the generation results of the investment strategy are not objective enough and the accuracy is not high enough, and the investment strategy is improved. Speed of generation as well as accuracy and objectivity.
  • step S2 includes:
  • the first evaluation factor is the self-attribute item used by the self-attribute tag to match the self-attribute information of the target product, according to the first evaluation factor
  • the scoring information of the target product is acquired, and when the scoring information is higher than a preset threshold, the target product is used as a first push product, and the first push product is input into the first push pool.
  • the server obtains the first evaluation factor of the self-attribute tag, and scores the stocks in the stock according to the first evaluation factor, specifically expressed as using the first evaluation factor
  • the factor scores the stock through its own attribute information of the stock, and generates corresponding scoring information, and then the server judges whether the scoring information is higher than a preset threshold, and if so, takes the stock as the second Push the product and add it to the first push pool, otherwise when the server judges whether the scoring information is lower than the preset threshold, the stock will not be used as the first stock;
  • the server judges that the first evaluation factor matched by the "high income” service tag input by the user is "return on net assets, gross profit margin”, then the server compares the net assets The rate of return and the gross profit rate score the stock according to the attribute information of the first pushed product, and generate corresponding scoring information, and then the server judges that the scoring information is higher than a preset threshold, and then takes the stock as the first Push the product and add it to the first pushing pool.
  • step S3 includes:
  • the server obtains the second evaluation factor of the external risk label, and then the server adds keywords contained in the second evaluation factor to the external risk information of each stock Matching is performed to obtain the number of successful matching matches, and then the server judges that when the number of successful matchings is higher than a preset threshold, the stock is used as the second push product, and the second push product is added to the In the second push pool, on the contrary, when the server judges that the number of successful matches is lower than the preset threshold, the stock will not be used as the second push product;
  • the server judges that the second evaluation factor of the external risk label of "bankruptcy risk” input by the user is the keyword “bankruptcy” and the keyword “executed”, then the server sends the The keywords “bankruptcy” and “executed” are sent to the external risk information of the stock for matching, and the two keywords “bankruptcy” and “executed” in the external risk information of each stock are obtained After the service judges that the number of occurrences is higher than the preset threshold, the stock is used as the second push product, and the second push product is added to the second push pool.
  • the acquiring the scoring information of the target product according to the first evaluation factor includes:
  • the server obtains the self-attribute value of the stock that matches the first evaluation factor, and discretizes the self-attribute value to obtain the discretization of the self-attribute value data, and then the server performs a validation test based on the discretized data to obtain the score of the first evaluation factor, and then the server adds the scores of the first evaluation factor to obtain the first Scoring information of evaluation factors;
  • the server obtains the return on equity value and the value of the gross profit rate corresponding to the return on equity and the gross profit rate in the stock's own attribute information, and then the server will Discretize the ROE value and the gross profit rate value to obtain the discrete value of the net asset value and the discrete value of the gross profit rate, and then the server validates the discrete value of the net asset value and the discrete value of the gross profit rate Test, so as to obtain the first evaluation factor score of the discrete value of the net assets and the first evaluation factor score of the discrete value of the gross profit rate, and then the server adds the two first evaluation factor scores to obtain The scoring information.
  • the discretization of the self-attribute value is performed to obtain discretized data
  • the validity test is performed on the discretized data, so as to obtain the first evaluation factor score, including:
  • the preset time range is the first 5 days, and the preset scoring rule is that if it is valid, 1 point will be added.
  • the preset rule is: the data of the day with the lowest ranking and the data of the day with the highest ranking The difference is less than the preset value;
  • the server obtains the data of the first 5 days in the self-attribute information of each of the stocks, thereby obtaining the training time period and the training attribute data, and discretizing the training attribute data to obtain historical discretization data , and then the server judges the ranking information of the historical discretization data corresponding to the first evaluation factor on each day in the training period, and judges the lowest ranking day of the historical discretization data in the ranking information
  • the difference between the data and the data on the day of the highest ranking is less than the preset value, and if so, it is judged that the first evaluation factor is valid, and 1 point is added to the corresponding factor score, and finally the factor score is output;
  • the preset time range, the preset scoring rules and the preset rules can be set by the user, which is not limited in this application.
  • the preset value is the fifth;
  • the server obtains the first 5 days of the self attribute information of the stock as a training period, and at the same time, the server obtains the specific data of the first 5 days of the self attribute information of the stock as the attribute data for training, and the training Discretization is performed with attribute data to obtain historical discretization data, and then the server obtains the discrete values of historical net assets and discrete values of historical gross profit margins (that is, historical discretized data) in the training period every day in the training period.
  • the server obtains the discrete values of historical net assets and discrete values of historical gross profit margins (that is, historical discretized data) in the training period every day in the training period.
  • the matching of the second evaluation factor based on the external risk information of each of the stocks is carried out to obtain the number of successful matches between the words in the external risk information of each of the stocks and the second evaluation factor, including :
  • the server obtains the keyword "bankruptcy” and the keyword “executed” of the second evaluation factor of the "bankruptcy risk” external risk label, and then the server obtains research report information and news information There are corresponding stocks of the "bankruptcy” keyword and the "executed” keyword, and the occurrence times of the "bankruptcy” keyword and the “executed” keyword in the research report information and the news information are obtained, And take the number of occurrences as the number of successful matches.
  • step S1 it also includes:
  • the server obtains the historical stock purchases of the user, and judges whether the self-attribute information of the historically purchased stocks is consistent with the first evaluation factor (net asset value) in the "high income" self-attribute tag. rate of return, gross profit margin), and whether the external risk information of the historically purchased stocks matches the second evaluation factor “bankruptcy” keyword and “executed” keyword that matches the external risk label of the “bankruptcy risk”, if If both the first evaluation factor and the second evaluation factor match successfully, the server pushes the historically purchased stocks to the user.
  • the first evaluation factor net asset value
  • rate of return gross profit margin
  • the server obtains the user's historical purchased stocks, and then the server obtains the own attribute information of the historically purchased stocks, and judges the net asset income in the historically purchased stocks' own attribute information rate and gross profit rate are higher than the preset value, and if so, it is judged that the historical purchased stock matches the self-attribute tag;
  • the server obtains the external risk information of the historically purchased stocks, and then the server adds the keywords contained in the second evaluation factor to the external risk information of each of the historically purchased stocks for matching, thereby obtaining Match the number of successful matches, and then the server judges whether the number of successful matches is within a preset threshold, and if so, judges that the historical purchased stock matches the external risk tag;
  • the server judges that when the historically purchased stock matches the self attribute tag and the external risk tag successfully, it pushes the historically purchased stock to the user.
  • the present application also provides an investment strategy generation device based on preset labels, including:
  • the preset label acquisition module 1 is used to acquire the preset label input by the user, the preset label includes the target product's own attribute label and the external risk label, and the self attribute label is the same as the target product's own attribute
  • An information-associated label, the external risk label is a label associated with the target product identity information
  • the application is applied in the field of investment
  • the target product may be a stock
  • the self-attribute label includes a label associated with its own attribute information
  • the self-attribute information includes its own financial scale information, self-identity information, etc.
  • self-financial scale information includes stock market value, price-earnings ratio, total share capital, dividend rate, etc.
  • self-identity information includes stock name, stock number, place of issue, company name, etc.
  • Self-attribute tags associated with self-attribute information include high income, high rate of return, steady return, high break rate, etc.;
  • the self-attribute tags are included in a preset self-attribute tag library, and each of the self-attribute tags is matched with a corresponding first evaluation factor.
  • the first evaluation factor of the "high-income" self-attribute tag is If the return on net assets and gross profit rate of the target product mentioned above, then only when the return on net assets and gross profit rate of the matched target product are higher than the preset value, can the attribute information of the target product and "high income" be judged Self-attribute tag matching;
  • the external risk tags are included in the preset external risk tag library, and the external risk tags include tags associated with non-financial information and identity information such as external market prices and news, such as non-financial information in public opinion information and research report information. , related information of identity information, news event reports and other related tags, which can specifically include: Hong Kong stock listing, shareholder divestment, bankruptcy risk, etc.
  • Each external risk tag is matched with a corresponding second evaluation factor.
  • the corresponding second evaluation factor is the keyword “bankruptcy” and “executed”. Keywords, when words such as “bankruptcy” and “executed” appear in the matching research reports and news of the target product, the external risk information of the target product can be determined to be related to the "emergence" keyword. Bankruptcy" self-attribute tag matching;
  • the first acquiring module 2 is configured to acquire a first push product matched with the self attribute tag, and input the first push product into a first push pool;
  • the first acquisition module 2 acquires the stocks that match the first evaluation factor (return on equity, gross profit rate) in the "high income" self-attribute tag, specifically expressed as the first acquisition module 2 acquires Stocks with ROE and gross profit margin higher than the preset value are used as the first push product, and the successfully matched stocks are gathered together to generate the first push pool;
  • first evaluation factor return on equity, gross profit rate
  • the first acquisition module 2 judges that the self-attribute label of "high income” input by the user matches the first evaluation factor as "return on net assets, gross profit margin", then the first acquisition module 2 Obtain 5 stocks whose return on equity and gross profit rate are higher than the preset value, and integrate the 5 stocks into the first push pool;
  • the second acquisition module 3 is configured to acquire a second push product matching the external risk label, and input the second push product into a second push pool;
  • the external risk information includes research report information and news information
  • the second acquisition module 3 acquires the "bankruptcy risk” external risk tag matching the second assessment factor is the keyword "bankruptcy” and "executed” keyword, then the second acquisition module 3 acquires the information contained in the research report information and news information The corresponding stocks of the "bankruptcy” keyword and the "executed” keyword, and gather the successfully matched stocks to generate the second push pool. It can be understood that the information contained in the external risk information can be provided by the user setting, this application does not limit it;
  • the second acquisition module 3 acquires that the second evaluation factor matched by the "bankruptcy risk” external risk label is the keyword “bankruptcy” and the keyword “executed”, then the second acquisition module 3 Obtain the 7 stocks with the keyword “bankruptcy” and “executed” in the research report information and news information, and integrate the 7 stocks into the second push pool.
  • a push module 4 configured to use the union of the first push pool and the second push pool as the product to be pushed;
  • the push module 4 merges the stocks contained in the first push pool and the second push pool to generate products to be pushed;
  • the push module 4 uses 5 stocks contained in the first push pool and 7 stocks in the second push pool, a total of 12 stocks, as products to be pushed;
  • the expected result information acquisition module 5 is used to input the products to be pushed into the pre-trained prediction model to obtain the expected result information of each of the products to be pushed;
  • the expected result information acquisition module 5 uses preset modeling to obtain the expected result information of each stock in the product to be pushed. It can be understood that the modeling of the preset modeling The method can be any one of xgb modeling or lightgbm modeling, which is not limited in this application;
  • the expected result information acquisition module 5 uses the lightgbm modeling set by the staff to calculate the expected result information of the 12 stocks in the product to be pushed;
  • An information push module 6 configured to push the expected result information and the product to be pushed to the user.
  • the information push module 6 calculates the expected result information of each stock in the product to be pushed, pushes the expected result information corresponding to the product to be pushed and each target product to be pushed to the user , so as to realize the generation of investment strategies and push them to corresponding users.
  • the information push module 6 pushes the 12 stocks in the product to be pushed and the expected result information of the 12 stocks to the customer's mobile phone, thereby realizing the push of the investment strategy.
  • the first acquisition module 2 includes:
  • a first evaluation factor acquiring unit configured to acquire a first evaluation factor of the self-attribute tag, where the first evaluation factor is a self-attribute item used by the self-attribute tag to match the self-attribute information of the target product;
  • a scoring information acquiring unit configured to acquire scoring information of the target product according to the first evaluation factor
  • a first product selection unit configured to use the target product as the first push product when the scoring information is higher than a preset threshold
  • a first product input unit configured to input the first pushed product into the first pushed pool.
  • the second acquisition module 3 includes:
  • a second evaluation factor obtaining unit configured to obtain a second evaluation factor of the external risk label, where the second evaluation factor is a keyword used by the external risk label to match the external risk information of the target product;
  • An information matching unit configured to match the second evaluation factor based on the external risk information of each target product, and obtain that the words in the external risk information of each target product are successfully matched with the second evaluation factor frequency;
  • a second product selection unit configured to select the target product corresponding to the external risk information whose matching success times are higher than a preset threshold, and use it as a second pushed product
  • a second product input unit configured to input the second push product into the second push pool.
  • the scoring information acquisition unit includes:
  • An attribute value calculation unit configured to obtain the own attribute value of the target product's own attribute item corresponding to the first evaluation factor
  • a score calculation unit configured to discretize the self-attribute value to obtain discretized data, and perform a validation test on the discretized data, so as to obtain the first evaluation factor score
  • the scoring information is obtained by adding up the scores of the first evaluation factors.
  • the discrete computing unit includes:
  • an attribute data acquisition unit configured to acquire historical attribute data of the target product
  • An attribute data intercepting unit configured to intercept from the historical attribute data according to a preset time range, to obtain training time periods and training attribute data
  • a discrete computing unit configured to discretize the attribute data for training to generate historical discretized data
  • a sorting information acquisition unit configured to acquire sorting information of historical discretized data in the training period
  • a validity evaluation unit configured to judge whether the first evaluation factor is valid according to preset rules and the ranking information
  • the score output unit is configured to output the corresponding score of the first evaluation factor according to a preset scoring rule for the first evaluation factor if yes.
  • the information matching unit includes:
  • An evaluation factor judging unit configured to judge whether the words in the external risk information of each target product include the second evaluation factor
  • a matching number acquisition unit configured to acquire the number of occurrences of the second evaluation factor in the external risk information of the target product if yes;
  • the success count acquisition unit is configured to use the occurrence count as the matching success count.
  • a product pushing module 7 is also included, including:
  • a historical product acquisition unit configured to acquire the user's historical purchase target product
  • a tag matching unit configured to determine whether the historical purchase target product matches the self attribute tag and the external risk tag
  • the target product pushing unit is configured to push the historically purchased target product to the user if yes.
  • the present application also provides a storage medium 100.
  • the computer-readable storage medium may be non-volatile or volatile.
  • a computer program 200 is stored in the storage medium 100. When it runs on a computer, the computer executes the preset label-based policy generation method described in the above embodiments, it includes: obtaining the preset label input by the user, the preset label including the target product's own attribute label and external risk label; obtaining and The first push product matching the self attribute label, and input the first push product into the first push pool; obtain the second push product matching the external risk label, and input the second push product into the first push pool; Two push pools; the union of the first push pool and the second push pool is used as the product to be pushed; the product to be pushed is input into a pre-trained prediction model to obtain the expected value of each product to be pushed Result information: pushing the expected result information and the product to be pushed to the user.
  • the present application also provides a computer device 300 containing instructions.
  • the computer device 300 executes the preset-based
  • the method for generating a label strategy includes: acquiring a preset label input by a user, the preset label including a target product's own attribute label and an external risk label; acquiring the first push product that matches the self attribute label, and The first push product is input into the first push pool; the second push product matching the external risk label is obtained, and the second push product is input into the second push pool; the first push pool is combined with the The union of the second push pool is used as the product to be pushed; the product to be pushed is input into a pre-trained prediction model to obtain the expected result information of each of the products to be pushed; The product is pushed to said user.
  • the greatest beneficial effect of the present application lies in: by obtaining the preset label input by the user, splitting the preset label into self-attribute label and external risk label, and obtaining the The product to be pushed that matches the external risk tag, and the revenue prediction of the product to be pushed is then pushed to the user.
  • the preset tag input by the user can be automatically split into its own attribute tag and the external risk tag.
  • ROM Read-Only Memory, read-only memory
  • RAM Random Access Memory, random access memory
  • EPROM Erasable Programmable Read-Only Memory
  • EEPROM Electrically Erasable Programmable Read-Only Memory, electrically erasable programmable read-only memory
  • flash memory magnetic card or optical card.
  • a readable medium includes any medium that stores or transmits information in a form readable by a device (eg, a computer).

Landscapes

  • Engineering & Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • Theoretical Computer Science (AREA)
  • Accounting & Taxation (AREA)
  • Finance (AREA)
  • General Physics & Mathematics (AREA)
  • Physics & Mathematics (AREA)
  • Strategic Management (AREA)
  • Economics (AREA)
  • Development Economics (AREA)
  • Marketing (AREA)
  • General Business, Economics & Management (AREA)
  • General Engineering & Computer Science (AREA)
  • Evolutionary Computation (AREA)
  • Software Systems (AREA)
  • Mathematical Physics (AREA)
  • Computing Systems (AREA)
  • Data Mining & Analysis (AREA)
  • Computational Linguistics (AREA)
  • Technology Law (AREA)
  • Artificial Intelligence (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

涉及人工智能领域，并具体提供了一种基于预设标签的策略生成方法、装置及存储介质，包括：取用户输入的预设标签，所述预设标签中包括目标产品的自身属性标签以及外部风险标签，获取与所述自身属性标签匹配的第一推送产品，并将所述第一推送产品输入第一推送池（S2），获取与所述外部风险标签匹配的第二推送产品，并将所述第二推送产品输入第二推送池（S3），将所述第一推送池与所述第二推送池的并集作为待推送产品（S4），将所述待推送产品输入预训练好的预测模型，以获取各所述待推送产品的预期结果信息（S5），将所述预期结果信息以及所述待推送产品推送给所述用户（S6），通过上述操作，实现了更精准的生成用户所需策略。

Description

基于预设标签的策略生成方法、装置及存储介质
本申请要求于2021年9月14日提交中国专利局、申请号为202111075330.3，发明名称为“基于预设标签的策略生成方法、装置及存储介质”的中国专利申请的优先权，其全部内容通过引用结合在本申请中。
技术领域
本申请涉及人工智能领域,尤其是基于预设标签的策略生成方法、装置及存储介质。
 背景技术
在现有的基于预设标签的策略生成方法中，量化投资策略的生成与实现是一个需要大量实验的过程，该过程涉及到因子挖掘、因子测试、因子组合、策略实验与策略监控等过程，整个策略实现的过程中，因子来源极其丰富，包含传统的财务指标、价量指标，技术面指标等不同频率的因子，也包含从文本（研报、舆情评论）等信息中挖掘加工出的指标，因此因子加工过程复杂，策略的生成过程也利用到了专家经验，机器学习、深度学习等方式，发明人意识到，整个实现过程面临巨大的搜索空间，同时在策略实现的过程中也要处理因子时变性与市场风格切换等诸多问题，导致需要考虑的因子过多，造成筛选效率下降同时投资策略的生成不够客观以及准确性不够高的问题。
 技术问题
本申请的目的为提供基于预设标签的策略生成方法、装置及存储介质，旨在解决现有技术中，策略的筛选需要考虑的因子过多，造成筛选效率下降同时投资策略的生成不够客观以及准确性不够高的问题。
 技术解决方案
本申请提供一种基于预设标签的策略生成方法，包括：获取用户输入的预设标签，所述预设标签中包括目标产品的自身属性标签以及外部风险标签；获取与所述自身属性标签匹配的第一推送产品，并将所述第一推送产品输入第一推送池；获取与所述外部风险标签匹配的第二推送产品，并将所述第二推送产品输入第二推送池；将所述第一推送池与所述第二推送池的并集作为待推送产品；将所述待推送产品输入预训练好的预测模型，以获取各所述待推送产品的预期结果信息；将所述预期结果信息以及所述待推送产品推送给所述用户。
本申请还提供了一种基于预设标签的投资策略生成装置，包括：预设标签获取模块，用于获取用户输入的预设标签，所述预设标签中包括目标产品的自身属性标签以及外部风险标签；第一获取模块，用于获取与所述自身属性标签匹配的第一推送产品，并将所述第一推送产品输入第一推送池；第二获取模块，用于获取与所述外部风险标签匹配的第二推送产品，并将所述第二推送产品输入第二推送池；推送模块，用于将所述第一推送池与所述第二推送池的并集作为待推送产品；预期结果信息获取模块，用于将所述待推送产品输入预训练好的预测模型，以获取各所述待推送产品的预期结果信息；信息推送模块，用于将所述预期结果信息以及所述待推送产品推送给所述用户。
本申请还提供了一种存储介质，存储介质中存储有计算机程序，当其在计算机上运行时，使得计算机执行以上实施例所描述的基于预设标签的策略生成方法，包括：获取用户输入的预设标签，所述预设标签中包括目标产品的自身属性标签以及外部风险标签；获取与所述自身属性标签匹配的第一推送产品，并将所述第一推送产品输入第一推送池；获取与所述外部风险标签匹配的第二推送产品，并将所述第二推送产品输入第二推送池；将所述第一推送池与所述第二推送池的并集作为待推送产品；将所述待推送产品输入预训练好的预测模型，以获取各所述待推送产品的预期结果信息；将所述预期结果信息以及所述待推送产品推送给所述用户。
本申请还提供了一种包含指令的计算机设备，当其在计算机设备上运行时，使得计算机设备通过其内部设置的处理器执行以上实施例所描述的基于预设标签的策略生成方法，包括：获取用户输入的预设标签，所述预设标签中包括目标产品的自身属性标签以及外部风险标签；获取与所述自身属性标签匹配的第一推送产品，并将所述第一推送产品输入第一推送池；获取与所述外部风险标签匹配的第二推送产品，并将所述第二推送产品输入第二推送池；将所述第一推送池与所述第二推送池的并集作为待推送产品；将所述待推送产品输入预训练好的预测模型，以获取各所述待推送产品的预期结果信息；将所述预期结果信息以及所述待推送产品推送给所述用户。
 有益效果
本申请的有益效果：通过获取用户输入的预设标签，并将所述预设标签拆分为自身属性标签以及外部风险标签，并获取与所述自身属性标签以及外部风险标签匹配的待推送产品，并将所述待推送产品进行收益预测之后再推送给用户的方式实现了可通过对用户输入的预设标签自动拆分为自身属性标签以及外部风险标签的方式实现了更精准生成策略同时加快了筛选效率，解决了现有技术当中现有的策略的生成需要考虑的因子过多，造成筛选效率下降同时策略的生成结果不够客观以及准确性不够高的问题，提升了策略生成的速度以及准确性和客观性。
 附图说明
图1为本申请的基于预设标签的策略生成方法的流程示意图；
图2为本申请的基于预设标签的策略生成装置的结构示意图；
图3为本申请的存储介质一实施例的结构框图；
图4为本申请的计算机设备一实施例的结构框图；
图中标号名称为：1-预设标签获取模块、2-标签拆分模块、3-第一获取模块、4-第二获取模块、5-推送模块、6-预期结果信息获取模块、7-信息推送模块、100-存储介质、200-计算机程序、300-计算机设备。
 本发明的最佳实施方式
应当理解，此处所描述的具体实施例仅仅用以解释本申请，并不用于限定本申请。
本技术领域技术人员可以理解，除非特意声明，这里使用的单数形式“一”、“一个”、“所述”和“该”也可包括复数形式。应该进一步理解的是，本申请的说明书中使用的措辞“包括”是指存在所述特征、整数、步骤、操作、元件和/或组件，但是并不排除存在或添加一个或多个其他特征、整数、步骤、操作、元件、组件和/或它们的组。这里使用的措辞“和/或”包括一个或更多个相关联的列出项的全部或任一单元和全部组合。
本申请实施例可以基于人工智能技术对相关的数据进行获取和处理。其中，人工智能（Artificial Intelligence，AI)是利用数字计算机或者数字计算机控制的机器模拟、延伸和扩展人的智能，感知环境、获取知识并使用知识获得最佳结果的理论、方法、技术及应用系统。
参考图1，本申请提供一种基于预设标签的策略生成方法，包括：
S1、获取用户输入的预设标签，所述预设标签中包括所述目标产品的自身属性标签以及外部风险标签，所述自身属性标签为与所述目标产品自身属性信息关联的标签，所述外部风险标签为与所述目标产品身份信息关联的标签；
在本申请的实施例中，本申请应用于投资领域，所述目标产品可以为股票，所述自身属性标签包括与其自身属性信息关联的标签，自身属性信息包括自身财务规模信息、自身身份信息等，自身财务规模信息包括股票市值、市盈率、总股本、股息率等；自身身份信息包括股票名称、股票编号、发行地、公司名称等。与自身属性信息关联的自身属性标签包括高收入、高回报率，稳健回报、高破发率等；
此外，所述自身属性标签包含在预设的自身属性标签库中，每个所述自身属性标签都匹配有对应的第一评估因子，如“高收入”自身属性标签的第一评估因子为所述目标产品的净资产收益率以及毛利率，则当匹配的所述目标产品的净资产收益率以及毛利率高于预设值时，才可判断该目标产品的自身属性信息与“高收入”自身属性标签匹配；
所述外部风险标签包含在预设的外部风险标签库中，所述外部风险标签包括与外部行情、消息等属于非财务信息和身份信息关联的标签，例如与舆论信息、研报信息中非财务、身份信息的相关信息、新闻事件报道等相关的标签，具体可以包括：港股上市、股东撤资、破产风险等。
每个外部风险标签都匹配有对应的第二评估因子，如当用户输入的所述外部风险标签为“破产风险”时，其对应的第二评估因子为“破产”关键字以及“被执行”关键字，当匹配的所述目标产品的所述研报及所述新闻中出现“破产”关键字以及“被执行”关键字等词语时，才可判定该目标产品的外部风险信息与“出现破产”自身属性标签匹配；
可以理解的是，所述自身属性标签以及所述外部风险标签的类型、所述自身属性标签的第一评估因子以及所述外部风险标签的第二评估因子可由用户自行设定，本申请不对此做限制；
此外，所述服务器可以是独立的服务器，也可以是提供云服务、云数据库、云计算、云函数、云存储、网络服务、云通信、中间件服务、域名服务、安全服务、内容分发网络(Content Delivery Network，CDN)、以及大数据和人工智能平台等基础云计算服务的云服务器。
S2、获取与所述自身属性标签匹配的第一推送产品，并将所述第一推送产品输入第一推送池；
在本申请一实施例中，所述服务器获取所述“高收入”自身属性标签中第一评估因子（净资产收益率、毛利率）匹配的目标产品，具体表现为所述服务器获取净资产收益率以及毛利率高于预设值的目标产品，作为为第一推送产品，并将成功匹配的所述目标产品集中起来输入所述第一推送池；
则在本申请一具体应用场景中，所述服务器判断所述用户输入的“高收入”自身属性标签匹配第一评估因子为“净资产收益率、毛利率”，则所述服务器获取净资产收益率以及毛利率高于预设值的5个股票，并将5个所述股票整合为第一推送池；
S3、获取与所述外部风险标签匹配的第二推送产品，并将所述第二推送产品输入第二推送池；
在本申请一实施例中，所述外部风险信息包含研报信息以及新闻信息；
所述服务器获取所述“破产风险”外部风险标签的第二评估因子为“破产”关键字以及“被执行”关键字，则所述服务器获取研报信息以及新闻信息中存在“破产”关键字以及“被执行”关键字的对应股票，并将成功匹配的所述股票集中起来生成所述第二推送池，可以理解的是，所述外部风险信息包含的信息可由用户自行设置，本申请不对其做限定；
则在本申请一具体应用场景中，所述服务器获取所述“破产风险”外部风险标签的第二评估因子为“破产”关键字以及“被执行”关键字，则所述服务器获取研报信息以及新闻信息中存在“破产”关键字以及“被执行”关键字的所述7个股票，并将7个所述股票整合为所述第二推送池。
S4、将所述第一推送池与所述第二推送池的并集作为待推送产品；
在本申请一实施例中，所述服务器将所述第一推送池以及所述第二推送池内包含的股票进行合并，从而生成待推送产品；
则在本申请一具体应用场景中，所述服务器将所述第一推送池内包含的5个股票以及所述第二推送池内的7个股票共12个股票作为待推送产品；
可以理解的是，当所述第一推送池内包含的股票与所述第二推送池内的包含的股票出现重合时（即所述第一推送池以及所述第二推送池的的交集），只添加一个重合股票到待推送产品中。
S5、将所述待推送产品输入预训练好的预测模型，以获取各所述待推送产品的预期结果信息；
在本申请一实施例中，所述服务器使用以预设收益预测模型来获取所述待推送产品中每一个股票的预期结果信息，可以理解的是，所述预设收益预测模型的建模方式可为xgb建模或lightgbm建模中的任一种，本申请对此不做限定；
在本申请一具体应用场景中，所述服务器使用工作人员设置的lightgbm建模来推算所述待推送产品中12个股票的预期结果信息；
S6、将所述预期结果信息以及所述待推送产品推送给所述用户。
在本申请一实施例中，所述服务器将所述待推送产品中各个股票的预期结果信息计算出来后，将所述待推送产品以及各个所述待推送产品对应的预期结果信息推送给用户，从而实现了投资策略的生成并推送给对应用户。
则在本申请一具体应用场景中，所述服务器将所述待推送产品中的12个股票以及该12个股票的预期结果信息推送给用户手机。
本申请的技术效果为：通过获取用户输入的预设标签，并将所述预设标签拆分为自身属性标签以及外部风险标签，并获取与所述自身属性标签以及外部风险标签匹配的待推送产品，并将所述待推送产品进行收益预测之后再推送给用户的方式实现了可通过对用户输入的预设标签自动拆分为自身属性标签以及外部风险标签的方式实现了更精准生成投资策略同时加快了筛选效率，解决了现有技术当中现有的投资策略的生成需要考虑的因子过多，造成筛选效率下降同时投资策略的生成结果不够客观以及准确性不够高的问题，提升了投资策略生成的速度以及准确性和客观性。
进一步地，所述步骤S2，包括：
S31、获取所述自身属性标签的第一评估因子，所述第一评估因子为所述自身属性标签用于与所述目标产品的自身属性信息匹配的自身属性事项，根据所述第一评估因子获取所述目标产品的打分信息，当所述打分信息高于预设阈值时，将所述目标产品作为第一推送产品，将所述第一推送产品输入所述第一推送池。
在本申请一实施例中，所述服务器获取所述自身属性标签的第一评估因子，并根据所述第一评估因子对所述股票内的股票进行打分，具体表现为使用所述第一评估因子通过所述股票的自身属性信息对所述股票进行评分，并生成对应的打分信息，之后所述服务器判断所述打分信息是否高于预设阈值，若是，则将所述股票作为所述第一推送产品并添加到所述第一推送池内，反之当所述服务器判断所述打分信息是否低于预设阈值时，则不将所述股票作为第一股票；
在本申请一具体应用场景中，所述服务器判断所述用户输入的“高收入”务标签匹配的第一评估因子为“净资产收益率、毛利率”，则所述服务器将所述净资产收益率以及毛利率针对第一推送产品的自身属性信息对所述股票进行评分，并生成对应的打分信息，之后所述服务器判断所述打分信息高于预设阈值，则将所述股票作为第一推送产品，并添加到所述第一推送池内。
进一步地，所述步骤S3，包括：
S31、获取所述外部风险标签的第二评估因子，所述第二评估因子为所述外部风险标签用于与所述目标产品的外部风险信息匹配的关键词，基于各个所述目标产品的外部风险信息对所述第二评估因子进行匹配，得出各所述目标产品的外部风险信息中的词语与所述第二评估因子的匹配成功次数，选取所述匹配成功次数高于预设阈值的外部风险信息对应的所述目标产品，并将其作为第二推送产品，将所述第二推送产品输入所述第二推送池。
在本申请一实施例中，所述服务器获取所述外部风险标签的第二评估因子，之后所述服务器将所述第二评估因子内包含的关键字添加到各个所述股票的外部风险信息中进行匹配，从而得出匹配匹配成功次数，之后所述服务器判断当所述匹配成功次数高于预设阈值时，将所述股票作为第二推送产品，并将所述第二推送产品添加到所述第二推送池内，反之当所述服务器判断所述匹配成功次数低于预设阈值时，则不将所述股票作为第二推送产品；
在本申请一具体应用场景中，所述服务器判断所述用户输入的“破产风险”外部风险标签的第二评估因子为“破产”关键字以及“被执行”关键字，则述服务器将所述“破产”关键字以及“被执行”关键字发送到所述股票的外部风险信息中进行匹配，并得出各个所述股票的的外部风险信息中“破产”以及“被执行”两个关键字的出现次数，之后所述服务判断所述出现次数高于预设阈值，则将所述股票作为第二推送产品，并将所述第二推送产品添加到所述第二推送池内。
进一步地，所述根据所述第一评估因子获取所述目标产品的打分信息，包括：
S32、获取所述第一评估因子对应的所述目标产品的自身属性事项的自身属性数值，将所述自身属性数值进行离散化，得出离散化数据，并对所述离散化数据进行有效化测试，从而得出第一评估因子得分，将所述第一评估因子得分进行相加，从而得出所述打分信息；
在本申请一实施例中，所述服务器获取所述股票与所述第一评估因子匹配的自身属性数值，并将所述自身属性数值进行离散化，得出针对所述自身属性数值的离散化数据，之后所述服务器根据所述离散化数据进行有效化测试，从而得出第一评估因子的得分，之后所述服务器将所述第一评估因子得分进行相加，从而得出所述第一评估因子的打分信息；
在本申请一具体应用场景中，所述服务器获取所述股票的自身属性信息中与所述净资产收益率以及所述毛利率对应的净资产收益率数值以及毛利率数值，之后所述服务器将所述净资产收益率数值以及所述毛利率数值进行离散化，从而得到净资产离散数值以及毛利率离散数值，之后所述服务器对所述净资产离散数值以及所述毛利率离散数值进行有效化测试，从而得出所述净资产离散数值的第一评估因子得分以及毛利率离散数值的第一评估因子得分，之后所述服务器将两个所述第一评估因子得分进行相加，从而得出所述打分信息。
进一步地，所述将所述自身属性数值进行离散化，得出离散化数据，并对所述离散化数据进行有效化测试，从而得出第一评估因子得分，包括：
S33、获取所述目标产品的历史属性数据,根据预设时间范围从所述历史属性数据中进行截取，得到训练用时段以及训练用属性数据,将所述训练用属性数据进行离散化，生成历史离散化数据,获取所述训练用时段中历史离散化数据的排序信息,根据预设规则以及所述排序信息判断所述第一评估因子是否有效,若是，则将所述第一评估因子根据预设打分规则输出对应的所述第一评估因子得分。
在本申请一实施例中，所述预设时间范围为前5天，所述预设打分规则为若有效，则加1分，所述预设规则为：最低排名当天数据与最高排名当天数据的差值小于预设值；
则所述服务器获取各个所述股票的自身属性信息中前5天的数据，从而得出训练用时段以及训练用属性数据，并对所述训练用属性数据进行离散化，得出历史离散化数据，之后所述服务器判断所述第一评估因子对应的历史离散化数据在所述训练用时段中的各天的排序信息，并判断在所述排序信息中所述历史离散化数据的最低排名当天数据与最高排名当天数据的差值小于预设值，若是，则判断所述第一评估因子有效，并在所述对应因子得分中加1分，并最终输出因子得分；
可以理解的是，所述预设时间范围、所述预设打分规则以及所述预设规则可由用户自行设置，本申请不对其做限定。
在本申请一具体应用场景中，所述预设值为第5名；
则所述服务器获取所述股票的自身属性信息中前5天为训练用时段，同时所述服务器获取所述股票的自身属性信息前5天的具体数据作为训练用属性数据，并对所述训练用属性数据进行离散化，得出历史离散化数据，之后所述服务器获取所述训练用时段中所述历史净资产离散数值以及历史毛利率离散数值（即历史离散化数据）每天在所述训练用时段对应数值排名表内的排序结果，并判断在所述排序结果中的所述净资产离散数值以及所述毛利率离散数值在最低排名当天数据与最高排名当天的数据是否小于预设值；
此时，所述服务器判断所述净资产离散数值在所述排序结果中最低排名当天数据与最高排名当天数据的差值小于预设值，所述毛利率离散数值在所述排序结果中最低排名当天数据与最高排名当天数据的差值大于预设值，则所述服务器判断第一评估因子“净资产收益率”有效，并向所述对应因子得分中加1分，而第一评估因子“毛利率”无效，则不向所述对应因子得分中加1分，则本次因子得分的计算结果为0+1+0=1分。
进一步地，所述基于各个所述股票的外部风险信息对所述第二评估因子进行匹配，得出各所述股票的外部风险信息中的词语与所述第二评估因子的匹配成功次数，包括：
判断各所述股票的外部风险信息中的词语是否包含所述第二评估因子，若是，则获取所述第二评估因子在所述股票的外部风险信息中的出现次数，将所述出现次数作为所述匹配成功次数。
在本申请一实施例中，所述服务器获取所述“破产风险”外部风险标签的第二评估因子“破产”关键字以及“被执行”关键字，则所述服务器获取研报信息以及新闻信息中存在“破产”关键字以及“被执行”关键字的对应股票，并获取所述“破产”关键字以及“被执行”关键字在所述研报信息以及所述新闻信息内的出现次数，并将所述出现次数作为匹配成功次数。
进一步地，所述步骤S1之前，还包括：
获取用户的历史购买目标产品，判断所述历史购买目标产品是否与所述自身属性标签以及所述外部风险标签匹配，若是，则将所述历史购买目标产品推送给用户。
在本申请一实施例中，所述服务器获取所述用户的历史购买股票，并判断所述历史购买股票的自身属性信息是否与所述“高收入”自身属性标签中第一评估因子（净资产收益率、毛利率）匹配，以及所述历史购买股票的外部风险信息是否与所述“破产风险”外部风险标签匹配的第二评估因子“破产”关键字以及“被执行”关键字匹配，若所述第一评估因子以及所述第二评估因子均匹配成功，则所述服务器将所述历史购买股票推送给用户。
在本申请一具体应用场景中，所述服务器获取用户的历史购买股票，之后所述服务器获取所述历史购买股票的自身属性信息，并判断所述历史购买股票的自身属性信息中的净资产收益率以及毛利率是否高于预设值，若是则判断所述历史购买股票与所述自身属性标签匹配；
同时所述服务器获取所述历史购买股票的外部风险信息，之后所述服务器将所述第二评估因子内包含的关键字添加到各个所述历史购买股票的外部风险信息中进行匹配，从而得出匹配匹配成功次数，之后所述服务器判断当所述匹配成功次数是否于预设阈值，若是则判断所述历史购买股票与所述外部风险标签匹配；
之后所述服务器判断当所述历史购买股票与所述自身属性标签以及所述外部风险标签匹配成功时，将所述历史购买股票推送给用户。
参考图2，本申请还提供一种基于预设标签的投资策略生成装置，包括：
预设标签获取模块1，用于获取用户输入的预设标签，所述预设标签中包括所述目标产品的自身属性标签以及外部风险标签，所述自身属性标签为与所述目标产品自身属性信息关联的标签，所述外部风险标签为与所述目标产品身份信息关联的标签；
在本申请的实施例中，本申请应用于投资领域，所述目标产品可以为股票，所述自身属性标签包括与其自身属性信息关联的标签，自身属性信息包括自身财务规模信息、自身身份信息等，自身财务规模信息包括股票市值、市盈率、总股本、股息率等；自身身份信息包括股票名称、股票编号、发行地、公司名称等。与自身属性信息关联的自身属性标签包括高收入、高回报率，稳健回报、高破发率等；
此外，所述自身属性标签包含在预设的自身属性标签库中，每个所述自身属性标签都匹配有对应的第一评估因子，如“高收入”自身属性标签的第一评估因子为所述目标产品的净资产收益率以及毛利率，则当匹配的所述目标产品的净资产收益率以及毛利率高于预设值时，才可判断该目标产品的自身属性信息与“高收入”自身属性标签匹配；
所述外部风险标签包含在预设的外部风险标签库中，所述外部风险标签包括与外部行情、消息等属于非财务信息和身份信息关联的标签，例如与舆论信息、研报信息中非财务、身份信息的相关信息、新闻事件报道等相关的标签，具体可以包括：港股上市、股东撤资、破产风险等。
每个外部风险标签都匹配有对应的第二评估因子，如当用户输入的所述外部风险标签为“破产风险”时，其对应的第二评估因子为“破产”关键字以及“被执行”关键字，当匹配的所述目标产品的所述研报及所述新闻中出现“破产”关键字以及“被执行”关键字等词语时，才可判定该目标产品的外部风险信息与“出现破产”自身属性标签匹配；
第一获取模块2，用于获取与所述自身属性标签匹配的第一推送产品，并将所述第一推送产品输入第一推送池；
在本申请一实施例中，第一获取模块2获取所述“高收入”自身属性标签中第一评估因子（净资产收益率、毛利率）匹配的股票，具体表现为第一获取模块2获取净资产收益率以及毛利率高于预设值的股票，作为为第一推送产品，并将成功匹配的所述股票集中起来生成所述第一推送池；
则在本申请一具体应用场景中，第一获取模块2判断所述用户输入的“高收入”自身属性标签匹配第一评估因子为“净资产收益率、毛利率”，则第一获取模块2获取净资产收益率以及毛利率高于预设值的5个股票，并将5个所述股票整合为第一推送池；
第二获取模块3，用于获取与所述外部风险标签匹配的第二推送产品，并将所述第二推送产品输入第二推送池；
在本申请一实施例中，所述外部风险信息包含研报信息以及新闻信息；
第二获取模块3获取所述“破产风险”外部风险标签匹配的第二评估因子为“破产”关键字以及“被执行”关键字，则第二获取模块3获取研报信息以及新闻信息中存在“破产”关键字以及“被执行”关键字的对应股票，并将成功匹配的所述股票集中起来生成所述第二推送池，可以理解的是，所述外部风险信息包含的信息可由用户自行设置，本申请不对其做限定；
则在本申请一具体应用场景中，第二获取模块3获取所述“破产风险”外部风险标签匹配的第二评估因子为“破产”关键字以及“被执行”关键字，则第二获取模块3获取研报信息以及新闻信息中存在“破产”关键字以及“被执行”关键字的所述7个股票，并将7个所述股票整合为所述第二推送池。
推送模块4，用于将所述第一推送池与所述第二推送池的并集作为待推送产品；
在本申请一实施例中，推送模块4将所述第一推送池以及所述第二推送池内包含的股票进行合并，从而生成待推送产品；
则在本申请一具体应用场景中，推送模块4将所述第一推送池内包含的5个股票以及所述第二推送池内的7个股票共12个股票作为待推送产品；
可以理解的是，当所述第一推送池内包含的股票与所述第二推送池内的包含的股票出现重合时，只添加一个重合股票到待推送产品中。
预期结果信息获取模块5，用于将所述待推送产品输入预训练好的预测模型，以获取各所述待推送产品的预期结果信息；
在本申请一实施例中，预期结果信息获取模块5使用以预设建模来获取所述待推送产品中每一个股票的预期结果信息，可以理解的是，所述预设建模的建模方式可为xgb建模或lightgbm建模中的任一种，本申请对此不做限定；
则在本申请一具体应用场景中，预期结果信息获取模块5使用工作人员设置的lightgbm建模来推算所述待推送产品中12个股票的预期结果信息；
信息推送模块6，用于将所述预期结果信息以及所述待推送产品推送给所述用户。
在本申请一实施例中，信息推送模块6将所述待推送产品中各个股票的预期结果信息计算出来后，将所述待推送产品以及各个所述推送目标产品对应的预期结果信息推送给用户，从而实现了投资策略的生成并推送给对应用户。
则在本申请一具体应用场景中，信息推送模块6将所述待推送产品中的12个股票以及该12个股票的预期结果信息推送给客户手机，从而实现投资策略的推送。
进一步地，第一获取模块2，包括：
第一评估因子获取单元，用于获取所述自身属性标签的第一评估因子，所述第一评估因子为所述自身属性标签用于与所述目标产品的自身属性信息匹配的自身属性事项；
打分信息获取单元，用于根据所述第一评估因子获取所述目标产品的打分信息；
第一产品选取单元，用于当所述打分信息高于预设阈值时，将所述目标产品作为第一推送产品；
第一产品输入单元，用于将所述第一推送产品输入所述第一推送池。
进一步地，第二获取模块3，包括：
第二评估因子获取单元，用于获取所述外部风险标签的第二评估因子，所述第二评估因子为所述外部风险标签用于与所述目标产品的外部风险信息匹配的关键词；
信息匹配单元，用于基于各个所述目标产品的外部风险信息对所述第二评估因子进行匹配，得出各所述目标产品的外部风险信息中的词语与所述第二评估因子的匹配成功次数；
第二产品选取单元，用于选取所述匹配成功次数高于预设阈值的外部风险信息对应的所述目标产品，并将其作为第二推送产品；
第二产品输入单元，用于将所述第二推送产品输入所述第二推送池。
进一步地，打分信息获取单元，包括：
属性数值计算单元，用于获取所述第一评估因子对应的所述目标产品的自身属性事项的自身属性数值；
得分计算单元，用于将所述自身属性数值进行离散化，得出离散化数据，并对所述离散化数据进行有效化测试，从而得出第一评估因子得分；
将所述第一评估因子得分进行相加，从而得出所述打分信息。
进一步地，离散计算单元，包括：
属性数据获取单元，用于获取所述目标产品的历史属性数据；
属性数据截取单元，用于根据预设时间范围从所述历史属性数据中进行截取，得到训练用时段以及训练用属性数据；
离散计算单元，用于将所述训练用属性数据进行离散化，生成历史离散化数据；
排序信息获取单元，用于获取所述训练用时段中历史离散化数据的排序信息；
有效性评估单元，用于根据预设规则以及所述排序信息判断所述第一评估因子是否有效；
得分输出单元，用于若是，则将所述第一评估因子根据预设打分规则输出对应的所述第一评估因子得分。
进一步地，信息匹配单元，包括：
评估因子判断单元，用于判断各所述目标产品的外部风险信息中的词语是否包含所述第二评估因子；
匹配次数获取单元，用于若是，则获取所述第二评估因子在所述目标产品的外部风险信息中的出现次数；
成功次数获取单元，用于将所述出现次数作为匹配成功次数。
进一步地，还包括产品推送模块7，包括：
历史产品获取单元，用于获取用户的历史购买目标产品；
标签匹配单元，用于判断所述历史购买目标产品是否与所述自身属性标签以及所述外部风险标签匹配；
目标产品推送单元，用于若是，则将所述历史购买目标产品推送给用户。
参考图3本申请还提供了一种存储介质100，所述计算机可读存储介质可以是非易失性，也可以是易失性，存储介质100中存储有计算机程序200，当其在计算机上运行时，使得计算机执行以上实施例所描述的基于预设标签的策略生成方法，包括：获取用户输入的预设标签，所述预设标签中包括目标产品的自身属性标签以及外部风险标签；获取与所述自身属性标签匹配的第一推送产品，并将所述第一推送产品输入第一推送池；获取与所述外部风险标签匹配的第二推送产品，并将所述第二推送产品输入第二推送池；将所述第一推送池与所述第二推送池的并集作为待推送产品；将所述待推送产品输入预训练好的预测模型，以获取各所述待推送产品的预期结果信息；将所述预期结果信息以及所述待推送产品推送给所述用户。
参考图4，本申请还提供了一种包含指令的计算机设备300，当其在计算机设备300上运行时，使得计算机设备300通过其内部设置的处理器400执行以上实施例所描述的基于预设标签的策略生成方法，包括：获取用户输入的预设标签，所述预设标签中包括目标产品的自身属性标签以及外部风险标签；获取与所述自身属性标签匹配的第一推送产品，并将所述第一推送产品输入第一推送池；获取与所述外部风险标签匹配的第二推送产品，并将所述第二推送产品输入第二推送池；将所述第一推送池与所述第二推送池的并集作为待推送产品；将所述待推送产品输入预训练好的预测模型，以获取各所述待推送产品的预期结果信息；将所述预期结果信息以及所述待推送产品推送给所述用户。
综合上述实施例可知，本申请最大的有益效果在于：通过获取用户输入的预设标签，并将所述预设标签拆分为自身属性标签以及外部风险标签，并获取与所述自身属性标签以及外部风险标签匹配的待推送产品，并将所述待推送产品进行收益预测之后再推送给用户的方式实现了可通过对用户输入的预设标签自动拆分为自身属性标签以及外部风险标签的方式实现了更精准生成投资策略同时加快了筛选效率，解决了现有技术当中现有的投资策略的生成需要考虑的因子过多，造成筛选效率下降同时投资策略的生成结果不够客观以及准确性不够高的问题，提升了投资策略生成的速度以及准确性和客观性。
本领域技术人员可以理解，本申请所述的智能设备的操作方法和上述所涉及用于执行本申请中所述方法中的一项或多项的设备。这些设备可以为所需的目的而专门设计和制造，或者也可以包括通用计算机中的已知设备。这些设备具有存储在其内的计算机程序或应用程序，这些计算机程序选择性地激活或重构。这样的计算机程序可以被存储在设备(例如，计算机)可读介质中或者存储在适于存储电子指令并分别耦联到总线的任何类型的介质中，所述计算机可读介质包括但不限于任何类型的盘(包括软盘、硬盘、光盘、CD-ROM、和磁光盘)、ROM(Read-Only Memory，只读存储器)、RAM(Random Access Memory，随机存储器)、EPROM(Erasable Programmable Read-Only Memory，可擦写可编程只读存储器)、EEPROM(Electrically Erasable Programmable Read-Only Memory，电可擦可编程只读存储器)、闪存、磁性卡片或光线卡片。也就是，可读介质包括由设备(例如，计算机)以能够读的形式存储或传输信息的任何介质。
以上所述仅为本申请的优选实施例，并非因此限制本申请的专利范围，凡是利用本申请说明书及附图内容所作的等效结构或等效流程变换，或直接或间接运用在其他相关的技术领域，均同理包括在本申请的专利保护范围内。

Claims (20)

  1. 一种基于预设标签的策略生成方法，其中，包括：
    获取用户输入的预设标签，所述预设标签中包括目标产品的自身属性标签以及外部风险标签；
    获取与所述自身属性标签匹配的第一推送产品，并将所述第一推送产品输入第一推送池；
    获取与所述外部风险标签匹配的第二推送产品，并将所述第二推送产品输入第二推送池；
    将所述第一推送池与所述第二推送池的并集作为待推送产品；
    将所述待推送产品输入预训练好的预测模型，以获取各所述待推送产品的预期结果信息；
    将所述预期结果信息以及所述待推送产品推送给所述用户。
  2. 如权利要求1所述的基于预设标签的策略生成方法，其中，所述获取与所述自身属性标签匹配的第一推送产品，并将所述第一推送产品输入第一推送池，包括：
    获取所述自身属性标签的第一评估因子，所述第一评估因子为所述自身属性标签用于与所述目标产品的自身属性信息匹配的自身属性事项；
    根据所述第一评估因子获取所述目标产品的打分信息；
    当所述打分信息高于预设阈值时，将所述目标产品作为第一推送产品；
    将所述第一推送产品输入所述第一推送池。
  3. 如权利要求1所述的基于预设标签的策略生成方法，其中，所述获取与所述外部风险标签匹配的第二推送产品，并将所述第二推送产品输入第二推送池，包括：
    获取所述外部风险标签的第二评估因子，所述第二评估因子为所述外部风险标签用于与所述目标产品的外部风险信息匹配的关键词；
    基于各个所述目标产品的外部风险信息对所述第二评估因子进行匹配，得出各所述目标产品的外部风险信息中的词语与所述第二评估因子的匹配成功次数；
    选取所述匹配成功次数高于预设阈值的外部风险信息对应的所述目标产品，并将其作为第二推送产品；
    将所述第二推送产品输入所述第二推送池。
  4. 如权利要求2所述的基于预设标签的策略生成方法，其中，所述根据所述第一评估因子获取所述目标产品的打分信息，包括：
    获取所述第一评估因子对应的所述目标产品的自身属性事项的自身属性数值；
    将所述自身属性数值进行离散化，得出离散化数据，并对所述离散化数据进行有效化测试，从而得出第一评估因子得分；
    将所述第一评估因子得分进行相加，从而得出所述打分信息。
  5. 如权利要求4所述的基于预设标签的策略生成方法，其中，所述将所述自身属性数值进行离散化，得出离散化数据，并对所述离散化数据进行有效化测试，从而得出第一评估因子得分，包括：
    获取所述目标产品的历史属性数据；
    根据预设时间范围从所述历史属性数据中进行截取，得到训练用时段以及训练用属性数据；
    将所述训练用属性数据进行离散化，生成历史离散化数据；
    获取所述训练用时段中历史离散化数据的排序信息；
    根据预设规则以及所述排序信息判断所述第一评估因子是否有效；
    若是，则将所述第一评估因子根据预设打分规则输出对应的所述第一评估因子得分。
  6. 如权利要求3所述的基于预设标签的策略生成方法，其中，所述基于各个所述目标产品的外部风险信息对所述第二评估因子进行匹配，得出各所述目标产品的外部风险信息中的词语与所述第二评估因子的匹配成功次数，包括：
    判断各所述目标产品的外部风险信息中的词语是否包含所述第二评估因子；
    若是，则获取所述第二评估因子在所述目标产品的外部风险信息中的出现次数；
    将所述出现次数作为匹配成功次数。
  7. 如权利要求1所述的基于预设标签的策略生成方法，其中，所述获取预设标签之前，还包括：
    获取用户的历史购买目标产品；
    判断所述历史购买目标产品是否与所述自身属性标签以及所述外部风险标签匹配；
    若是，则将所述历史购买目标产品推送给用户。
  8. 一种基于预设标签的投资策略生成装置，其中，包括：
    预设标签获取模块，用于获取用户输入的预设标签，所述预设标签中包括目标产品的自身属性标签以及外部风险标签；
    第一获取模块，用于获取与所述自身属性标签匹配的第一推送产品，并将所述第一推送产品输入第一推送池；
    第二获取模块，用于获取与所述外部风险标签匹配的第二推送产品，并将所述第二推送产品输入第二推送池；
    推送模块，用于将所述第一推送池与所述第二推送池的并集作为待推送产品；
    预期结果信息获取模块，用于将所述待推送产品输入预训练好的预测模型，以获取各所述待推送产品的预期结果信息；
    信息推送模块，用于将所述预期结果信息以及所述待推送产品推送给所述用户。
  9. 一种存储介质，其特征在于，其为计算机可读的存储介质，其上存储有计算机程序，所述计算机程序被执行时实现一种基于预设标签的策略生成方法的步骤；
    其中，所述基于预设标签的策略生成方法包括:
    获取用户输入的预设标签，所述预设标签中包括目标产品的自身属性标签以及外部风险标签；
    获取与所述自身属性标签匹配的第一推送产品，并将所述第一推送产品输入第一推送池；
    获取与所述外部风险标签匹配的第二推送产品，并将所述第二推送产品输入第二推送池；
    将所述第一推送池与所述第二推送池的并集作为待推送产品；
    将所述待推送产品输入预训练好的预测模型，以获取各所述待推送产品的预期结果信息；
    将所述预期结果信息以及所述待推送产品推送给所述用户。
  10. 如权利要求9所述的存储介质，其中，所述获取与所述自身属性标签匹配的第一推送产品，并将所述第一推送产品输入第一推送池，包括：
    获取所述自身属性标签的第一评估因子，所述第一评估因子为所述自身属性标签用于与所述目标产品的自身属性信息匹配的自身属性事项；
    根据所述第一评估因子获取所述目标产品的打分信息；
    当所述打分信息高于预设阈值时，将所述目标产品作为第一推送产品；
    将所述第一推送产品输入所述第一推送池。
  11. 如权利要求9所述的存储介质，其中，所述获取与所述外部风险标签匹配的第二推送产品，并将所述第二推送产品输入第二推送池，包括：
    获取所述外部风险标签的第二评估因子，所述第二评估因子为所述外部风险标签用于与所述目标产品的外部风险信息匹配的关键词；
    基于各个所述目标产品的外部风险信息对所述第二评估因子进行匹配，得出各所述目标产品的外部风险信息中的词语与所述第二评估因子的匹配成功次数；
    选取所述匹配成功次数高于预设阈值的外部风险信息对应的所述目标产品，并将其作为第二推送产品；
    将所述第二推送产品输入所述第二推送池。
  12. 如权利要求10所述的存储介质，其中，所述根据所述第一评估因子获取所述目标产品的打分信息，包括：
    获取所述第一评估因子对应的所述目标产品的自身属性事项的自身属性数值；
    将所述自身属性数值进行离散化，得出离散化数据，并对所述离散化数据进行有效化测试，从而得出第一评估因子得分；
    将所述第一评估因子得分进行相加，从而得出所述打分信息。
  13. 如权利要求12所述的存储介质，其中，所述将所述自身属性数值进行离散化，得出离散化数据，并对所述离散化数据进行有效化测试，从而得出第一评估因子得分，包括：
    获取所述目标产品的历史属性数据；
    根据预设时间范围从所述历史属性数据中进行截取，得到训练用时段以及训练用属性数据；
    将所述训练用属性数据进行离散化，生成历史离散化数据；
    获取所述训练用时段中历史离散化数据的排序信息；
    根据预设规则以及所述排序信息判断所述第一评估因子是否有效；
    若是，则将所述第一评估因子根据预设打分规则输出对应的所述第一评估因子得分。
  14. 如权利要求11所述的存储介质，其中，所述基于各个所述目标产品的外部风险信息对所述第二评估因子进行匹配，得出各所述目标产品的外部风险信息中的词语与所述第二评估因子的匹配成功次数，包括：
    判断各所述目标产品的外部风险信息中的词语是否包含所述第二评估因子；
    若是，则获取所述第二评估因子在所述目标产品的外部风险信息中的出现次数；
    将所述出现次数作为匹配成功次数。
  15. 一种计算机设备，其中，其包括处理器、存储器及存储于所述存储器上并可在所述处理器上运行的计算机程序，所述处理器执行所述计算机程序时实现一种基于预设标签的策略生成方法的步骤；
    其中，所述基于预设标签的策略生成包括：
    获取用户输入的预设标签，所述预设标签中包括目标产品的自身属性标签以及外部风险标签；
    获取与所述自身属性标签匹配的第一推送产品，并将所述第一推送产品输入第一推送池；
    获取与所述外部风险标签匹配的第二推送产品，并将所述第二推送产品输入第二推送池；
    将所述第一推送池与所述第二推送池的并集作为待推送产品；
    将所述待推送产品输入预训练好的预测模型，以获取各所述待推送产品的预期结果信息；
    将所述预期结果信息以及所述待推送产品推送给所述用户。
  16. 如权利要求15所述的计算机设备，其中，所述获取与所述自身属性标签匹配的第一推送产品，并将所述第一推送产品输入第一推送池，包括：
    获取所述自身属性标签的第一评估因子，所述第一评估因子为所述自身属性标签用于与所述目标产品的自身属性信息匹配的自身属性事项；
    根据所述第一评估因子获取所述目标产品的打分信息；
    当所述打分信息高于预设阈值时，将所述目标产品作为第一推送产品；
    将所述第一推送产品输入所述第一推送池。
  17. 如权利要求15所述的计算机设备，其中，所述获取与所述外部风险标签匹配的第二推送产品，并将所述第二推送产品输入第二推送池，包括：
    获取所述外部风险标签的第二评估因子，所述第二评估因子为所述外部风险标签用于与所述目标产品的外部风险信息匹配的关键词；
    基于各个所述目标产品的外部风险信息对所述第二评估因子进行匹配，得出各所述目标产品的外部风险信息中的词语与所述第二评估因子的匹配成功次数；
    选取所述匹配成功次数高于预设阈值的外部风险信息对应的所述目标产品，并将其作为第二推送产品；
    将所述第二推送产品输入所述第二推送池。
  18. 如权利要求16所述的计算机设备，其中，所述根据所述第一评估因子获取所述目标产品的打分信息，包括：
    获取所述第一评估因子对应的所述目标产品的自身属性事项的自身属性数值；
    将所述自身属性数值进行离散化，得出离散化数据，并对所述离散化数据进行有效化测试，从而得出第一评估因子得分；
    将所述第一评估因子得分进行相加，从而得出所述打分信息。
  19. 如权利要求18所述的计算机设备，其中，所述将所述自身属性数值进行离散化，得出离散化数据，并对所述离散化数据进行有效化测试，从而得出第一评估因子得分，包括：
    获取所述目标产品的历史属性数据；
    根据预设时间范围从所述历史属性数据中进行截取，得到训练用时段以及训练用属性数据；
    将所述训练用属性数据进行离散化，生成历史离散化数据；
    获取所述训练用时段中历史离散化数据的排序信息；
    根据预设规则以及所述排序信息判断所述第一评估因子是否有效；
    若是，则将所述第一评估因子根据预设打分规则输出对应的所述第一评估因子得分。
  20. 如权利要求17所述的计算机设备，其中，所述基于各个所述目标产品的外部风险信息对所述第二评估因子进行匹配，得出各所述目标产品的外部风险信息中的词语与所述第二评估因子的匹配成功次数，包括：
    判断各所述目标产品的外部风险信息中的词语是否包含所述第二评估因子；
    若是，则获取所述第二评估因子在所述目标产品的外部风险信息中的出现次数；
    将所述出现次数作为匹配成功次数。
PCT/CN2022/072364 2021-09-14 2022-01-17 基于预设标签的策略生成方法、装置及存储介质 WO2023040155A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202111075330.3 2021-09-14
CN202111075330.3A CN113781246B (zh) 2021-09-14 2021-09-14 基于预设标签的策略生成方法、装置及存储介质

Publications (1)

Publication Number Publication Date
WO2023040155A1 true WO2023040155A1 (zh) 2023-03-23

Family

ID=78843729

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2022/072364 WO2023040155A1 (zh) 2021-09-14 2022-01-17 基于预设标签的策略生成方法、装置及存储介质

Country Status (2)

Country Link
CN (1) CN113781246B (zh)
WO (1) WO2023040155A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117407514A (zh) * 2023-11-28 2024-01-16 星环信息科技(上海)股份有限公司 一种解决计划生成方法、装置、设备及存储介质

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113781246B (zh) * 2021-09-14 2024-03-05 平安科技（深圳）有限公司 基于预设标签的策略生成方法、装置及存储介质

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8694405B2 (en) * 2010-01-21 2014-04-08 Visual Trader Systems, S.L. System and method of visual representation of stock exchange transactions
CN109146680A (zh) * 2018-07-20 2019-01-04 徐温杰 选股方法及系统
CN109636501A (zh) * 2018-10-30 2019-04-16 深圳壹账通智能科技有限公司 产品数据处理方法、装置、计算机设备及计算机可读存储介质
CN112991026A (zh) * 2021-05-08 2021-06-18 明品云(北京)数据科技有限公司 一种商品推荐方法、系统、设备及计算机可读存储介质
CN113781246A (zh) * 2021-09-14 2021-12-10 平安科技（深圳）有限公司 基于预设标签的策略生成方法、装置及存储介质

Family Cites Families (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR20160073251A (ko) * 2014-12-16 2016-06-24 삼성생명보험주식회사 금융상품 운용 방법, 장치 및 컴퓨터 프로그램
CN106993030A (zh) * 2017-03-22 2017-07-28 北京百度网讯科技有限公司 基于人工智能的信息推送方法和装置
CN107798607A (zh) * 2017-07-25 2018-03-13 上海壹账通金融科技有限公司 资产配置策略获取方法、装置、计算机设备和存储介质
US11068897B2 (en) * 2017-12-21 2021-07-20 Paypal, Inc. Completing risk analysis using push communications
CN108337316A (zh) * 2018-02-08 2018-07-27 平安科技（深圳）有限公司 信息推送方法、装置、计算机设备及存储介质
CN109615106A (zh) * 2018-10-16 2019-04-12 平安科技（深圳）有限公司 股票收益推送方法、装置、存储介质和服务器
CN109615487A (zh) * 2019-01-04 2019-04-12 平安科技（深圳）有限公司 基于用户行为的产品推荐方法、装置、设备及存储介质
CN111861569B (zh) * 2020-07-23 2024-02-02 中国工商银行股份有限公司 产品信息推荐方法及装置
CN112184302A (zh) * 2020-09-25 2021-01-05 中国建设银行股份有限公司 一种产品推荐方法、装置、规则引擎及存储介质
CN112330404A (zh) * 2020-11-10 2021-02-05 广发证券股份有限公司 数据处理方法和装置、服务器及存储介质

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8694405B2 (en) * 2010-01-21 2014-04-08 Visual Trader Systems, S.L. System and method of visual representation of stock exchange transactions
CN109146680A (zh) * 2018-07-20 2019-01-04 徐温杰 选股方法及系统
CN109636501A (zh) * 2018-10-30 2019-04-16 深圳壹账通智能科技有限公司 产品数据处理方法、装置、计算机设备及计算机可读存储介质
CN112991026A (zh) * 2021-05-08 2021-06-18 明品云(北京)数据科技有限公司 一种商品推荐方法、系统、设备及计算机可读存储介质
CN113781246A (zh) * 2021-09-14 2021-12-10 平安科技（深圳）有限公司 基于预设标签的策略生成方法、装置及存储介质

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117407514A (zh) * 2023-11-28 2024-01-16 星环信息科技(上海)股份有限公司 一种解决计划生成方法、装置、设备及存储介质

Also Published As

Publication number Publication date
CN113781246A (zh) 2021-12-10
CN113781246B (zh) 2024-03-05

Similar Documents

Publication Publication Date Title
US11475143B2 (en) Sensitive data classification
US11334635B2 (en) Domain specific natural language understanding of customer intent in self-help
CN108846520B (zh) 贷款逾期预测方法、装置以及计算机可读存储介质
US10387437B2 (en) Query rewriting using session information
CN102708131B (zh) 将消费者自动分类到微细分中
US20170140464A1 (en) Method and apparatus for evaluating relevance of keyword to asset price
WO2023040155A1 (zh) 基于预设标签的策略生成方法、装置及存储介质
US20110258054A1 (en) Automatic Generation of Bid Phrases for Online Advertising
KR20150036113A (ko) 검색 결과를 랭킹하는 방법과 시스템 및 검색 결과 랭킹을 최적화하는 방법과 시스템
WO2022089190A1 (zh) 产品推荐方法、装置、电子设备及可读存储介质
CN113283675A (zh) 指标数据分析方法、装置、设备及存储介质
CN111986792A (zh) 医疗机构评分方法、装置、设备及存储介质
WO2022007798A1 (zh) 数据显示方法、装置、终端设备及存储介质
CN105405051A (zh) 金融事件预测方法和装置
Jha et al. Reputation systems: Evaluating reputation among all good sellers
US20220188876A1 (en) Advertising method and apparatus for generating advertising strategy
CN107464019A (zh) 一种金融事件预警方法
CN113269179A (zh) 数据处理方法、装置、设备及存储介质
CN112395401A (zh) 自适应负样本对采样方法、装置、电子设备及存储介质
CN113343700B (zh) 数据处理方法、装置、设备及存储介质
JP2021012420A (ja) 顧客管理システム、及び顧客管理方法
CN113240325B (zh) 数据处理方法、装置、设备及存储介质
JP2016197332A (ja) 情報処理システム、情報処理方法、およびコンピュータプログラム
CN117480521A (zh) 用于确定标识符记录的强度的方法、设备和非暂时性计算机可读存储介质
CN113849618A (zh) 基于知识图谱的策略确定方法、装置、电子设备及介质

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 22868536

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 22868536

Country of ref document: EP

Kind code of ref document: A1