CN111833098A - Information prediction method, storage medium and electronic device - Google Patents
Information prediction method, storage medium and electronic device Download PDFInfo
- Publication number
- CN111833098A CN111833098A CN202010581008.7A CN202010581008A CN111833098A CN 111833098 A CN111833098 A CN 111833098A CN 202010581008 A CN202010581008 A CN 202010581008A CN 111833098 A CN111833098 A CN 111833098A
- Authority
- CN
- China
- Prior art keywords
- sales
- type
- activity
- historical
- keyword
- Prior art date
- Legal status (The legal status 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 status listed.)
- Withdrawn
Links
Images
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
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0201—Market modelling; Market analysis; Collecting market data
- G06Q30/0202—Market predictions or forecasting for commercial activities
-
- 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
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0207—Discounts or incentives, e.g. coupons or rebates
Landscapes
- Business, Economics & Management (AREA)
- Strategic Management (AREA)
- Engineering & Computer Science (AREA)
- Accounting & Taxation (AREA)
- Development Economics (AREA)
- Finance (AREA)
- Entrepreneurship & Innovation (AREA)
- Economics (AREA)
- Game Theory and Decision Science (AREA)
- Marketing (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The embodiment of the application discloses an information prediction method, a storage medium and an electronic device. The method comprises the following steps: if the keywords of the commodity do not appear in the historical data of the target type sales activities planned to be released, determining the released sales activities with the keywords to appear, and obtaining at least two reference types of sales activities; acquiring historical index information of the keywords under the sales activities of each reference type; and predicting the index information of the keywords under the target type sales activities by utilizing the historical index information to obtain a prediction result.
Description
Technical Field
Embodiments of the present disclosure relate to the field of information processing, and more particularly, to an information prediction method, a storage medium, and an electronic device.
Background
In shopping promotion activities such as product promotion and the like carried out on shopping websites, the keyword index performance in a future period is predicted by the historical index performance (such as ROI, click rate and the like) of the thrown keywords, so that intuitive guide information is provided for the selection of the keywords, and the accuracy of the later-stage throwing is facilitated.
In order to realize targeted delivery, the information such as the type of sales activities, brands and categories to be selected in a period to be predicted in the future can be obtained, corresponding index prediction is carried out on keywords used by the obtained information in the historical product delivery process, keywords with poor performance are screened out according to the predicted values of the indexes, and the keywords with good performance are used as keywords used for delivery operation.
When an index for predicting a keyword of a commodity under a certain sales type is executed, if the keyword is not used under the sales type in the historical delivery, the accuracy of the prediction result for the keyword is low.
Disclosure of Invention
In order to solve any one of the above technical problems, embodiments of the present application provide an information prediction method, a storage medium, and an electronic device.
To achieve the purpose of the embodiment of the present application, an embodiment of the present application provides an information prediction method, including:
if the keywords of the commodity do not appear in the historical data of the target type sales activities planned to be released, determining the released sales activities with the keywords to appear, and obtaining at least two reference types of sales activities;
acquiring historical index information of the keywords under the sales activities of each reference type;
and predicting the index information of the keywords under the target type sales activities by utilizing the historical index information to obtain a prediction result.
A storage medium having a computer program stored therein, wherein the computer program is arranged to perform the method as described above when executed.
An electronic device comprising a memory having a computer program stored therein and a processor arranged to execute the computer program to perform the method as described above.
One of the above technical solutions has the following advantages or beneficial effects:
if the keyword of the commodity does not appear in historical data of the planned-to-be-released target-type sales activity, determining the released sales activity with the keyword, obtaining at least two reference-type sales activities, obtaining historical index information of the keyword under each reference-type sales activity, predicting the index information of the keyword under the target-type sales activity by using the historical index information to obtain a prediction result, achieving the purpose of completing prediction of the index of the keyword under the target-type sales activity by using the index information of the keyword under the reference-type sales activity, and improving the accuracy of index prediction of the keyword aiming at different activity types.
Additional features and advantages of the embodiments of the application will be set forth in the description which follows, and in part will be obvious from the description, or may be learned by the practice of the embodiments of the application. The objectives and other advantages of the embodiments of the application may be realized and attained by the structure particularly pointed out in the written description and claims hereof as well as the appended drawings.
Drawings
The accompanying drawings are included to provide a further understanding of the embodiments of the present application and are incorporated in and constitute a part of this specification, illustrate embodiments of the present application and together with the examples of the embodiments of the present application do not constitute a limitation of the embodiments of the present application.
Fig. 1 is a flowchart of an information prediction method provided in an embodiment of the present application;
fig. 2 is a flowchart of a method for predicting index information of a keyword according to an embodiment of the present disclosure.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present application more apparent, the embodiments of the present application will be described in detail below with reference to the accompanying drawings. It should be noted that, in the embodiments of the present application, features in the embodiments and the examples may be arbitrarily combined with each other without conflict.
The problem to be solved by the embodiment of the application is the cold start problem of index prediction of the keywords aiming at different activity scenes under the E-market scene. The cold start is defined as: given an activity type, such as short, daily, etc., a performance prediction is made for a keyword, and the keyword has never been delivered under that activity type, i.e., there is no historical performance record for that activity.
In the process of implementing the present application, the inventor conducts technical analysis on the related art, and finds that the related art has at least the following problems, including:
by way of specific application example, keywords used when the sale activity of laundry detergent under a certain brand is a big promotion are predicted. If the keyword "laundry detergent" is not used in the sales activity of "big promotions" but only appears in the sales activity of "daily", the click rate performance of the keyword during future big promotions will be inaccurate if to be predicted.
In the related art, a neighbor mean method is adopted, and the predicted index performs the same regardless of the activity type during the prediction. The different types of activities in the actual scene, such as whether the activities are short or short, daily activities, etc., may affect the index performance of the keywords.
In the related art, the following problems exist by adopting a time sequence algorithm:
disadvantage 1: the amount of data in the actual scene is not large enough
The time sequence algorithm is established on a machine learning model, and one disadvantage of the machine learning model is that the effect depends on the size of data volume to a great extent, and if the data volume is too small, the model can hardly capture the correlation among the keywords.
And (2) disadvantage: the features that can be extracted from the keywords in the actual scene are limited
When selecting the features of the time-series algorithm, the algorithm is greatly inclined to the performance of the neighbor mean value because the selectable features of the keywords are few and most of the features depend on the historical performance of the algorithm, such as the historical rolling mean value, the historical statistical variable and the like.
Based on the above analysis, it can be seen that, when index prediction is performed on a keyword lacking historical data of an activity type for different activity types, one method of the related art is that information of future activity types is not used, so that a prediction result is single, and the other method is that index prediction accuracy of the keyword cannot be improved due to data volume limitation. Therefore, how to improve the accuracy of keyword index prediction for different activity types under the condition of rare data is an urgent problem to be solved.
Based on the above analysis, the embodiments of the present application provide the following solutions, including:
fig. 1 is a flowchart of an information prediction method according to an embodiment of the present application. As shown in fig. 1, the method shown in fig. 1 includes:
in an exemplary embodiment, the type of sales activity is daily sales, a promotion on a particular date (e.g., holidays per year, promotional days set by a shopping website), or special offers.
In one exemplary embodiment, the commodities are divided according to brands of the commodities and categories of the commodities, so that the commodities are subdivided, and support is provided for subsequent accurate prediction.
Taking the commodity C in the type B under the brand A as an example for explanation, if a certain keyword d is planned to be put in the special price activity, the keyword d does not appear in the special price activity, but the keyword d appears when the commodity C is sold daily and promoted greatly, the sales activity of the reference type is determined to be daily sales and promotion.
102, acquiring historical index information of the keywords under each reference type of sales activities;
in an exemplary embodiment, the historical index information under the sales activities of different reference types is used as the input information for predicting the future index information under the sales activities of the target type, and the prediction operation can be completed by means of the historical information under the sales activities of the reference model of the keyword on the premise that the historical data is too little to be suitable for the machine learning model.
And 103, predicting index information of the keywords under the target type sales activities by using the historical index information to obtain a prediction result.
In an exemplary embodiment, index information corresponding to sales activities of different reference types is calculated, and compared with prediction by adopting an averaging method in the related art, the influence of time factors on prediction results can be overcome, and differences under different activity types can be reflected.
According to the method provided by the embodiment of the application, if the keyword of the commodity does not appear in historical data of the planned-to-be-released target-type sales activity, the released sales activity with the keyword is determined to appear, at least two reference-type sales activities are obtained, historical index information of the keyword under each reference-type sales activity is obtained, index information of the keyword under the target-type sales activity is predicted by utilizing the historical index information, a prediction result is obtained, the purpose of completing prediction of the index of the keyword under the target-type sales activity by means of the index information of the keyword under the reference-type sales activity is achieved, and the accuracy of index prediction of the keyword aiming at different activity types is improved.
The method provided by the embodiments of the present application is explained as follows:
in an exemplary embodiment, the predicting the index information of the keyword under the target type of sales activity by using the historical index information to obtain a prediction result includes:
determining a brand and a category to which the commodity belongs;
according to the historical index information of the keywords under the reference type sales activities, historical performance information corresponding to the brands and the categories of the keywords under the reference type sales activities is calculated;
and obtaining the performance information of the keyword under the target type sales activities according to the historical performance information of the keyword under the reference type sales activities.
Under the reference type sales activity, historical index information of commodities with the same brand and the same category of the planned commodities is obtained, prediction operation of the brand and the category based on the prediction granularity is realized, and the aggregated data has statistical significance and is not influenced by data quantity.
In an exemplary embodiment, the performance information of the keyword under the target type of sales activity is obtained by the following method, including:
obtaining a weight of the sales activity of each reference type;
and weighting the historical performance information of each target activity by using the obtained weight to obtain the performance information of the keyword under the target type of sales activity.
By determining the weight of each reference type of sales activity and then performing a weighting operation, the accuracy of the prediction can be improved.
In an exemplary embodiment, the weight of each reference type of sales activity is obtained by:
calculating the ratio of index performance information between any two reference types of sales activities;
and constructing a transposed matrix by utilizing the ratio of the index performance information between the sales activities of any two reference types, and taking the transposed matrix as the weight of each reference type of sales activity.
By pre-estimating a transpose matrix of the activity type, the content of the matrix is an index performance ratio between the activity types under the given brand and category, and then predicting the performance of the keyword under the activity type which does not appear through the transpose matrix.
The difference of different activity types can be reflected more accurately by determining the transpose matrix of the activity types, and support is provided for more accurate prediction.
In an exemplary embodiment, the ratio of the index performance information may be obtained by any one of the following methods, including:
acquiring the historical mean value of the keyword in each reference type of sales activity, and calculating the ratio of the historical mean values of any two reference types of sales activities;
and calculating the ratio of the historical mean values of any two reference types of sales activities through a preset machine learning model.
The process of estimating the ratio is to calculate a Transformation Matrix, and the value in the Transformation Matrix may be estimated using a historical mean ratio, or using a machine learning model, such as a neural network, a tree model, or using a statistical method.
In an exemplary embodiment, the performance information of the keyword under the target type of sales activity is further calculated according to a preset normalization coefficient.
The difference of the sales activities of different reference types can be further highlighted through the normalization coefficient, and support is provided for the performance information under the sales activities of the target types to be accurately predicted subsequently.
In an exemplary embodiment, the normalization factor is determined based on the number of days released for each reference type of sales activity.
In an exemplary embodiment, the target type of performance information b under the sales activitykThe calculation expression of (a) is as follows:
where k denotes the number of the sales activity of the target type, i and j denote the numbers of the sales activities of the reference type, rijA ratio of index performance information, m, representing a reference type of sales activity i to a reference type of sales activity jjThe number of days the keyword has been released in the sales activity j is represented, m represents the total number of days the keyword has been released in the sales activity, wherein i and j are positive integers, and the value is less than or equal to n, wherein n is a positive integer.
And determining a normalization coefficient by using the number of days released by each reference type of sales activity so as to determine the proportion of each reference type of sales activity and improve the prediction accuracy.
The method provided by the embodiments of the present application is explained as follows:
fig. 2 is a flowchart of a method for predicting index information of a keyword according to an embodiment of the present disclosure. As shown in fig. 2, the method shown in fig. 2 includes:
given a keyword name to be predicted, a target name, a time point t to be predicted, and an activity type k, brand, category, and the like for the time period.
And judging whether the keyword has historical index performance corresponding to the activity type k. If so, a preset algorithm is adopted for prediction. If not, executing the following steps 1 to 3, wherein:
and Step 1, estimating index performance ratio values among activity types under the brand categories to construct a transposed matrix, wherein the estimation methods are various, machine learning algorithms can be used, and a simpler averaging method (the part can be modularized) can also be used. The following is an example of an averaging method: the historical mean ratio of the two activity types is directly taken, and the granularity is (brand x category). That is, given an activity type i, j, the historical mean values pi, pj of the activity i, j under the brand category combination are calculated, respectively. The final ratio rij-pi/pj
Step 2, calculating the expression mean value of the keyword in each historical activity: a isiI is 1, …, n; wherein n is an integer; if not thrown under a certain activity, set to 0.
Step 3, assuming that the activity to be predicted is k, defining bkIs the predicted value of the index at k activity, where,
where mj/m represents the proportion of days that the keyword has been placed in j activity, which serves as a normalization.
The transposition matrix used by the method provided by the embodiment of the application is an averaging method, but is not limited to the averaging method, and can be estimated by using methods such as machine learning and the like. For example, characteristics are created according to brands, categories and performances in a certain past time window, and the characteristics are input to a machine learning algorithm to estimate activity type proportion.
According to the method provided by the embodiment of the application, under the condition of a small amount of data, the index performance in a certain activity type in the future can be estimated by estimating the index performance ratio between the activity types in real time and combining the historical performance of the keyword, so that the index prediction accuracy can be greatly improved. Compared with the prior mean value method, different predictions can be made for different activity types, and the problem of single prediction is solved. Compared with the performance of the index directly predicted by the machine learning algorithm, if the index is directly predicted by the machine learning algorithm, the predicted granularity is too fine, and if the data size is not large enough, the predicted result is not credible, and the noise is large. The invention is less affected by the data amount because the prediction granularity of the invention is thicker when predicting the transpose matrix, and the aggregated data has more statistical significance.
A storage medium having a computer program stored therein, wherein the computer program is arranged to perform the method of any of the above when executed.
An electronic device comprising a memory having a computer program stored therein and a processor arranged to run the computer program to perform the method of any of the above.
It will be understood by those of ordinary skill in the art that all or some of the steps of the methods, systems, functional modules/units in the devices disclosed above may be implemented as software, firmware, hardware, and suitable combinations thereof. In a hardware implementation, the division between functional modules/units mentioned in the above description does not necessarily correspond to the division of physical components; for example, one physical component may have multiple functions, or one function or step may be performed by several physical components in cooperation. Some or all of the components may be implemented as software executed by a processor, such as a digital signal processor or microprocessor, or as hardware, or as an integrated circuit, such as an application specific integrated circuit. Such software may be distributed on computer readable media, which may include computer storage media (or non-transitory media) and communication media (or transitory media). The term computer storage media includes volatile and nonvolatile, removable and non-removable media implemented in any method or technology for storage of information such as computer readable instructions, data structures, program modules or other data, as is well known to those of ordinary skill in the art. Computer storage media includes, but is not limited to, RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, Digital Versatile Disks (DVD) or other optical disk storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other medium which can be used to store the desired information and which can accessed by a computer. In addition, communication media typically embodies computer readable instructions, data structures, program modules or other data in a modulated data signal such as a carrier wave or other transport mechanism and includes any information delivery media as known to those skilled in the art.
Claims (10)
1. An information prediction method, comprising:
if the keywords of the commodity do not appear in the historical data of the target type sales activities planned to be released, determining the released sales activities with the keywords to appear, and obtaining at least two reference types of sales activities;
acquiring historical index information of the keywords under the sales activities of each reference type;
and predicting the index information of the keywords under the target type sales activities by utilizing the historical index information to obtain a prediction result.
2. The method according to claim 1, wherein the predicting index information of the keyword under the target type of sales activity by using the historical index information to obtain a prediction result comprises:
determining a brand and a category to which the commodity belongs;
according to the historical index information of the keywords under the reference type sales activities, historical performance information corresponding to the brands and the categories of the keywords under the reference type sales activities is calculated;
and obtaining the performance information of the keyword under the target type sales activities according to the historical performance information of the keyword under the reference type sales activities.
3. The method of claim 2, wherein the performance information of the keyword under the target type of sales activity is obtained by:
obtaining a weight of the sales activity of each reference type;
and weighting the historical performance information of each target activity by using the obtained weight to obtain the performance information of the keyword under the target type of sales activity.
4. The method of claim 3, wherein the weight for each reference type of sales activity is obtained by:
calculating the ratio of index performance information between any two reference types of sales activities;
and constructing a transposed matrix by utilizing the ratio of the index performance information between the sales activities of any two reference types, and taking the transposed matrix as the weight of each reference type of sales activity.
5. The method of claim 4, wherein the ratio of the index performance information is obtained by any one of the following methods, including:
acquiring the historical mean value of the keyword in each reference type of sales activity, and calculating the ratio of the historical mean values of any two reference types of sales activities;
and calculating the ratio of the historical mean values of any two reference types of sales activities through a preset machine learning model.
6. The method according to any one of claims 2 to 5, wherein the performance information of the keyword in the target type of sales activity is further calculated according to a preset normalization coefficient.
7. The method of claim 6, wherein the normalization factor is determined based on the number of days released for each reference type of sales activity.
8. The method of claim 7, wherein the performance information b for the targeted type of sales activitykThe calculation expression of (a) is as follows:
where k denotes the number of the sales activity of the target type, i and j denote the numbers of the sales activities of the reference type, rijA ratio of index performance information, m, representing a reference type of sales activity i to a reference type of sales activity jjThe number of days the keyword has been released in the sales activity j is represented, m represents the total number of days the keyword has been released in the sales activity, wherein i and j are positive integers, and the value is less than or equal to n, wherein n is a positive integer.
9. A storage medium, in which a computer program is stored, wherein the computer program is arranged to perform the method of any of claims 1 to 8 when executed.
10. An electronic device comprising a memory and a processor, wherein the memory has stored therein a computer program, and wherein the processor is arranged to execute the computer program to perform the method of any of claims 1 to 8.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010581008.7A CN111833098A (en) | 2020-06-23 | 2020-06-23 | Information prediction method, storage medium and electronic device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010581008.7A CN111833098A (en) | 2020-06-23 | 2020-06-23 | Information prediction method, storage medium and electronic device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111833098A true CN111833098A (en) | 2020-10-27 |
Family
ID=72898231
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010581008.7A Withdrawn CN111833098A (en) | 2020-06-23 | 2020-06-23 | Information prediction method, storage medium and electronic device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111833098A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113420202A (en) * | 2021-07-15 | 2021-09-21 | 上海明略人工智能(集团)有限公司 | Method and device for predicting keyword search times, electronic equipment and storage medium |
-
2020
- 2020-06-23 CN CN202010581008.7A patent/CN111833098A/en not_active Withdrawn
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113420202A (en) * | 2021-07-15 | 2021-09-21 | 上海明略人工智能(集团)有限公司 | Method and device for predicting keyword search times, electronic equipment and storage medium |
| CN113420202B (en) * | 2021-07-15 | 2024-04-02 | 上海明略人工智能(集团)有限公司 | Method and device for predicting keyword search times, electronic equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110033314B (en) | Advertisement data processing method and device | |
| CN110263242A (en) | Content recommendation method, device, computer readable storage medium and computer equipment | |
| CN110210882A (en) | Promote position matching process and device, promotion message methods of exhibiting and device | |
| US20120253945A1 (en) | Bid traffic estimation | |
| US20090222321A1 (en) | Prediction of future popularity of query terms | |
| CN112613938B (en) | Model training method and device and computer equipment | |
| CN107633257B (en) | Data quality evaluation method and device, computer readable storage medium and terminal | |
| CN111144874A (en) | Payment mode recommendation method, device and equipment | |
| US9324026B2 (en) | Hierarchical latent variable model estimation device, hierarchical latent variable model estimation method, supply amount prediction device, supply amount prediction method, and recording medium | |
| CN108182633A (en) | Loan data processing method, device, computer equipment and storage medium | |
| CN112423134A (en) | Video content recommendation method and device and computer equipment | |
| CN111833098A (en) | Information prediction method, storage medium and electronic device | |
| US20050165596A1 (en) | Method and apparatus for determining expected values in the presence of uncertainty | |
| CN115730125A (en) | Object identification method and device, computer equipment and storage medium | |
| CN112070564B (en) | Advertisement pulling method, device and system and electronic equipment | |
| CN111582448B (en) | Weight training method and device, computer equipment and storage medium | |
| CN110851708B (en) | Negative sample extraction method, device, computer equipment and storage medium | |
| CN115841345B (en) | Cross-border big data intelligent analysis method, system and storage medium | |
| CN110796379B (en) | Risk assessment method, device and equipment of business channel and storage medium | |
| CN109523296B (en) | User behavior probability analysis method and device, electronic equipment and storage medium | |
| Aliaj et al. | Nowcasting inflation with Lasso‐regularized vector autoregressions and mixed frequency data | |
| CN108154377B (en) | Advertisement cheating prediction method and device | |
| CN116739665A (en) | Information delivery method and device, electronic equipment and storage medium | |
| CN114329160A (en) | Searching method, searching device, terminal and storage medium | |
| CN114298825A (en) | Method and device for extremely evaluating repayment volume |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WW01 | Invention patent application withdrawn after publication |
Application publication date: 20201027 |
|
| WW01 | Invention patent application withdrawn after publication |