CN115329760A - Promotion keyword simulation screening method and system - Google Patents
Promotion keyword simulation screening method and system Download PDFInfo
- Publication number
- CN115329760A CN115329760A CN202211268303.2A CN202211268303A CN115329760A CN 115329760 A CN115329760 A CN 115329760A CN 202211268303 A CN202211268303 A CN 202211268303A CN 115329760 A CN115329760 A CN 115329760A
- Authority
- CN
- China
- Prior art keywords
- keyword
- score
- keywords
- search
- sub
- 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.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 39
- 238000012216 screening Methods 0.000 title claims abstract description 32
- 238000004088 simulation Methods 0.000 title claims description 3
- 238000012163 sequencing technique Methods 0.000 claims abstract description 15
- 238000004364 calculation method Methods 0.000 claims abstract description 11
- 230000001737 promoting effect Effects 0.000 claims description 16
- 230000011218 segmentation Effects 0.000 claims description 12
- 238000006243 chemical reaction Methods 0.000 claims description 9
- 239000002131 composite material Substances 0.000 claims description 4
- 238000010276 construction Methods 0.000 claims description 3
- 239000000126 substance Substances 0.000 claims description 2
- 230000000694 effects Effects 0.000 abstract description 5
- 238000005457 optimization Methods 0.000 abstract description 2
- 238000011156 evaluation Methods 0.000 description 6
- 238000004422 calculation algorithm Methods 0.000 description 3
- 238000004891 communication Methods 0.000 description 3
- 230000008901 benefit Effects 0.000 description 2
- 239000000284 extract Substances 0.000 description 2
- 238000013459 approach Methods 0.000 description 1
- 230000003796 beauty Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 239000002537 cosmetic Substances 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000012549 training Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/279—Recognition of textual entities
- G06F40/284—Lexical analysis, e.g. tokenisation or collocates
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/33—Querying
- G06F16/332—Query formulation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/258—Heading extraction; Automatic titling; Numbering
-
- 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]
Abstract
The invention discloses a method and a system for simulating and screening promotion keywords, which belong to the technical field of online shop promotion, and comprise the following steps: acquiring a search keyword set of a user group through big data; constructing a search ranking database according to the search keyword set; recommending a candidate keyword set for a target user according to given promoted article information; obtaining a plurality of selected keywords screened by a target user, calculating product correlation scores, creative correlation scores and product quality scores of the selected keywords, and performing weighted summation calculation to obtain the keyword quality scores of the selected keywords; and sequencing all the selected keywords according to the quality scores of the keywords from high to low, and displaying the sequencing results in a list form. The method and the system can intuitively simulate the promotion degree of the keywords through numerical sorting such as quality score, estimated ranking and the like, avoid blind word selection when a target user promotes online stores, and improve the promotion optimization effect.
Description
Technical Field
The invention belongs to the technical field of online shop popularization, and particularly relates to a method and a system for simulating and screening popularization keywords.
Background
Under the promotion of the internet economy tide, with continuous breakthrough and innovation of the e-commerce industry, the e-commerce also continuously changes our lives and becomes an essential part of our daily life. Compared with the past, the flow bonus period is past, the fine operation period is reached at the present stage, how to obtain more flows and obtain higher marketing conversion rate is a difficult problem for various large electronic commerce enterprises, online store promotion talents are urgently needed in the market, however, huge flow promotion cost needs to be paid for the cultivation of the online store promotion talents, students hardly train promotion skills with the help of real environment, simple classroom explanation is abstract, and the effect is very little.
In the aspect of screening the promotion keywords, after characteristics of candidate promotion keywords are extracted, a part of the prior art uses a trained keyword screening model to predict the high-quality promotion keywords, a traditional screening mode which simply depends on a fixed threshold and has strong regularity is replaced, keywords which are not effective in a promotion system can also be predicted, and the accuracy of screening the high-quality promotion keywords is improved. This approach, however, relies on a large number of training samples and does not provide a benefit to the trainee who needs to master the promotional skills.
In the preheating putting time period, the keywords can be screened according to the quality scores of the keywords, and the keywords are dynamically added and deleted to realize dynamic adjustment, but the quality scores depend on the actual preheating putting effect, so that the method is not practical for students who are difficult to develop electronic commerce operation popularization skills by means of a real environment.
The online shop popularization is a core skill which any e-commerce practitioner must master, but at present, the cultivation of the popularization skill only stays in a theoretical explanation level, and theoretical application and application results cannot be verified.
Therefore, in order to avoid blindly screening the popularization keywords, a method or a system is needed to help the trainees to perform keyword simulation popularization rehearsal, and the popularization keyword selection capability is gradually improved.
Disclosure of Invention
In view of the above, the invention provides a method and a system for simulating and screening promotion keywords, which are used for solving the problem of blindly screening the promotion keywords during online shop promotion.
The invention discloses a method for simulating and screening promotion keywords in a first aspect, which comprises the following steps:
acquiring a search keyword set of a user group through big data;
constructing a search ranking database according to the search keyword set;
recommending a candidate keyword set for a target user according to given promoted article information;
obtaining a plurality of selected keywords screened from the candidate keyword set by the target user, and calculating the product relevance score, the creative relevance score and the product quality score of the selected keywords;
weighting and summing the product relevance score, the creative relevance score and the product quality score to obtain the keyword quality score of each selected keyword;
and sequencing all the selected keywords according to the quality scores of the keywords from high to low, and displaying the sequencing results in a list form.
On the basis of the above technical solution, preferably, the constructing a search ranking database according to the search keyword set specifically includes:
acquiring search keywords of an nth user in a user group, and extracting core keywords of the search keywords, wherein N =1,2, \8230, and N is the total number of the users in the user group;
extracting product related words and keyword related words of the search keywords according to the core keywords;
forming the core keywords, the product related words and the keyword related words into nth search words;
obtaining the nth search result information according to the probability of each product in the search result of the nth search word;
repeating the steps until the Nth search result information is obtained;
and collecting and summarizing all search result information to form a search ranking database.
On the basis of the above technical solution, preferably, the search result corresponding to each search keyword in the search ranking database carries multiple attribute information, including a display amount, a click amount, a conversion amount, a click rate, a conversion rate, a rating score, a total rating score, a rating score and a product rating score.
On the basis of the above technical solution, preferably, after obtaining a plurality of selected keywords screened from the candidate keyword set by the target user, before calculating the product relevance score, the creative relevance score and the product quality score of the selected keywords, the method further includes:
performing word segmentation on the selected keyword to obtain a sub-keyword sequence;
calculating the closeness score of the selected keyword according to the sub-keyword sequence and the given promoted article information;
calculating the relevance score of each sub keyword according to the closeness score of the selected keyword;
and integrating the relevance scores of the sub-keywords, calculating the comprehensive score of the selected keyword, and taking the comprehensive score of the selected keyword as the keyword relevance score of the selected keyword.
On the basis of the above technical solution, preferably, the calculating the closeness score of the selected keyword according to the sub-keyword sequence and the given promoted item information specifically includes:
finding out the position of the sub-keyword sequence in the given promotion article information, calculating the distance between the sub-keywords according to the word number of the sub-keywords after word segmentation in the given promotion article information, and scoring the closenessρThe calculation formula of (2) is as follows:
wherein MaxForScore is a preset maximum scoring distance, max is the maximum distance value in the sub-keywords smaller than the maximum scoring distance, remove is the number of words in the sub-keywords that are not related to the given promotional item information or whose distance exceeds the maximum scoring distance, and remove Max is the upper limit of remove.
On the basis of the foregoing technical solution, preferably, the calculating the relevance score of each sub-keyword according to the closeness score of the selected keyword specifically includes:
wherein the content of the first and second substances,
j =1,2, \8230forthe relevance score of the jth sub-keyword, t +1 is the total number of the sub-keywords of the selected keyword, f is the total word frequency of the sub-keywords of the selected keyword in the given promotion article information; PVScore is the browsing volume score, i.e. the ratio of the browsing volume of the currently selected keyword to the average value of the browsing volume in the search ranking database, CTRScore is the click rate score, i.e. the ratio of the click rate of the currently selected keyword to the average value of the click rate in the search ranking database,
。
on the basis of the above technical solution, preferably, the integrating the relevance scores of the sub-keywords and calculating the comprehensive score of the selected keyword specifically includes:
let sequence { x 1 ,x 2 ,…,x t， x t+1 And f, integrating the relevance scores of the sub-keywords by adopting the following recursive formula, wherein the relevance score is the sequence of the relevance scores of the sub-keyword sequence, and t +1 is the total number of the sub-keywords:
wherein, { S 1 ,S 2 ,…,S t The integrated score sequence is used as the score sequence,
a composite score for the selected keyword.
On the basis of the above technical solution, preferably, the calculating the product relevance score, the creative relevance score and the product quality score of the selected keyword specifically includes:
searching all product related words according to given popularization article information to form a product related word set, searching related words of the currently selected keywords from the product related word set to form a keyword related word set; judging whether the keyword related word set and the selected keyword have an inclusion relationship, and calculating the product relevance score of the selected keyword according to the keyword relevance score of the selected keyword;
segmenting the given popularization article information, judging whether the selected keyword and the sub-keyword after segmenting the given popularization article information have an inclusion or included relationship, calculating a keyword correlation score of the sub-keyword with the inclusion or included relationship, and calculating an creative correlation score according to the keyword correlation score of the sub-keyword;
and comparing the product attribute information corresponding to the selected keyword with each average attribute value of all products under the category to which the current product belongs, calculating the score of each attribute based on the set product attribute weight ratio, and taking the sum of the scores of the attributes as the product quality score of the selected keyword.
On the basis of the foregoing technical solution, preferably, the comparing the product attribute information corresponding to the selected keyword with each average attribute value of all products in the category to which the current product belongs, and calculating the score of each attribute based on the set product attribute weight ratio specifically includes:
let the weight ratio of the attribute i be W i ，
The attribute value of the attribute i in the current product attribute information is R i The average attribute value of each of all products in the category to which the current product belongs is R i_0 Then the score T of each attribute i i The calculation formula of (2) is as follows:
where i =1,2, \8230, M, M is the total number of attributes.
In a second aspect of the present invention, a system for simulating and screening a promoted keyword is disclosed, the system comprising:
a database construction module: the system comprises a search keyword set, a search ranking database and a database server, wherein the search keyword set is used for acquiring a user group through big data, and the search ranking database is constructed according to the search keyword set;
the keyword screening module: the system is used for recommending a candidate keyword set for a target user according to given promoted article information; obtaining a plurality of selected keywords screened from a candidate keyword set by a target user;
the scoring and ranking module: for calculating a product relevance score, a creative relevance score, and a product quality score for the selected keyword; weighting and summing the product relevance score, the creative relevance score and the product quality score to obtain the keyword quality score of each selected keyword; and sequencing all the selected keywords from high to low according to the quality scores of the keywords, and displaying the sequencing result in a list form.
Compared with the prior art, the invention has the following beneficial effects:
1) The method constructs a search ranking database, performs traversal search ranking on the selected keywords of the target user, comprehensively calculates the keyword quality scores of the selected keywords by calculating the product relevance scores, the creative relevance scores and the product quality scores of the selected keywords, intuitively simulates the promotion degree of the keywords through numerical sorting such as the keyword quality scores and the estimated ranking, avoids blind word selection when the target user promotes online stores, and improves the promotion optimization effect;
2) According to the method, the selected keywords are segmented, the positions of the sub-keywords in the title or description of the given promoted article information are searched, the closeness score is calculated, the relevance score of each sub-keyword is calculated according to the closeness score, and the relevance degree between each sub-keyword in the selected keywords and the given promoted article information can be evaluated more accurately;
3) According to the method, the relevance scores of all the sub-keywords are integrated to calculate the comprehensive score of the selected keyword, the comprehensive score is used as the keyword relevance score of the selected keyword, the keyword relevance calculation accuracy can be improved, and the accuracy of popularization effect evaluation of the popularization keyword is further improved.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the drawings without creative efforts.
Fig. 1 is a flow chart of a method for simulating and screening a promotion keyword according to the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be obtained by a person skilled in the art without any inventive step based on the embodiments of the present invention, are within the scope of the present invention.
Referring to fig. 1, the present invention provides a method for simulating and screening a promoted keyword, the method including:
s1, acquiring a search keyword set of a user group through big data.
A user inputs a keyword in a platform search engine, and usually obtains a plurality of search results, and the rank of the search results is given a precedence, namely the search rank. Therefore, the search ranking refers to that the user inputs related links displayed on the home page by the keywords on the search engine, and the search engine performs sequencing display according to the popularity of the search keywords of the user.
Specifically, the method obtains the search keywords of the crowd users from a specific e-commerce search platform website (such as Taobao) through http and encryption protocol technologies such as HttpWebRequest, webResponse, servicePointManager and the like, obtains the search keyword set of the users by capturing, managing, processing and arranging the platform users through big data, and classifies the keywords, such as food, general merchandise, electric appliances, beauty cosmetics, women's clothing and the like.
And S2, constructing a search ranking database according to the search keyword set.
Assuming that the total number of users in the user group is N, N =1,2, \ 8230, N, the step S2 specifically includes the following sub-steps:
s21, obtaining the search keywords of the nth user in the user group, and extracting the core keywords of the search keywords.
And S22, extracting product related words and keyword related words of the search keywords according to the core keywords.
Specifically, the search keyword of the user may be a single keyword or a combination of multiple keywords, the product related words are searched from the search keyword set according to the search keyword of the nth user to form a product related word set, and in the process, the related words of the same kind of products searched by other users are searched from the search keyword set according to the core keyword, for example, if the core keyword is "shoe", the product related words may be "sports shoe", "casual shoe", and the like, and if the core keyword is "men's shoe", the product related words may be "breathable sports shoe men's shoe", "men's shoe new style", "men's shoe wrapped in mail", and the like.
And searching related words of the current search keywords from the product related word set according to the inclusion or included relation, performing semantic conversion to obtain semantic similar words, replacement words and expansion words, and forming the related words of the current search keywords, the semantic similar words, the replacement words and the expansion words into a keyword related word set.
Wherein, the searching logic for searching the related words of the current searching keyword from the product related word set is as follows: an inclusion or contained relationship. The inclusion or contained relation here refers to an inclusion or contained relation between the sub keyword, the search keyword, and the product related word after the search keyword is segmented. For example, if the current keyword is "new dress", the keyword "dress" is included in the "new dress", so the "dress" is put into the keyword related word set; the keyword 'summer new dress' includes 'new dress', so the 'summer new dress' puts into the keyword related words set.
Meanwhile, the related words of the keywords may also be semantic similar words, alternative words and extension words related to the keywords, for example, the related keywords of the "mobile phone" include "smart phone", "apple phone", "android phone", "wireless charging phone", "elderly mobile phone", and the like.
And S23, forming the core keywords, the product related words and the keyword related words into nth search words.
Specifically, the nth search word is a searchable phrase that is composed of a core keyword, a product related word, and a keyword related word according to the search keyword of the nth user.
And S24, obtaining the nth search result information according to the probability of each product in the search results of the nth search term.
And inputting the nth search word into the website of the e-commerce search platform to obtain the nth search result information, wherein the search result has a search ranking.
S25, repeating the steps S21-S24 until Nth search result information is obtained;
and S26, collecting and summarizing the 1 st search result information to the Nth search result information to form a search ranking database.
The product search results corresponding to each search keyword in the search ranking database carry a plurality of attribute information, including display amount, click rate, conversion rate, good evaluation number, total evaluation number, good evaluation rate, product evaluation and the like.
The search ranking database is used for searching the preliminary ranking of each keyword obtained through big data, has the advantages of massive data, complete information and high search efficiency, and provides technical support for subsequent keyword screening and keyword quality score calculation.
And S3, recommending a candidate keyword set for the target user according to the given promoted article information.
In order to train the popularization keyword selection capability of online store students, the method and the system give a piece of information of an article to be popularized as a topic, select the popularization keyword of the information of the article to be popularized as a task, allow the students to select the popularization keyword, and perform selected popularization keyword evaluation and optimal popularization keyword recommendation.
In order to better help a target user to screen keywords, the method extracts core keywords according to given promoted article information, extracts product related words and keyword related words of the promoted article information according to the core keywords, and screens the keywords with high relevance as a candidate keyword set for a student to screen.
Specifically, the way of screening the keywords with higher relevance is as follows:
calculating a keyword correlation score of each related word in the keyword related word set;
calculating a product relevance score of the product relevant words according to the keyword relevance score, wherein if the current product relevant words exist in the keyword relevant word set, the product relevance score is equal to the keyword relevance score, and if the current product relevant words do not exist in the keyword relevant word set, the product relevance score is equal to half of the maximum keyword relevance score in the keyword relevant word set;
arranging the related words of the keywords according to the sequence of the relevance scores of the keywords from high to low;
arranging the product related words in the order of the product relevance scores from high to low;
screening keywords with keyword relevance scores and product relevance scores higher than a third preset threshold value to form a candidate keyword set;
and presenting the candidate keywords in the candidate keyword set in a list form according to the sequence of the product relevance scores from high to low, and recommending the candidate keywords to the target user in a tag selection mode for screening by the target user.
And S4, acquiring a plurality of selected keywords screened from the candidate keyword set by the target user, and calculating the product relevance score, the creative relevance score and the product quality score of the selected keywords.
Step S4 specifically includes the following sub-steps:
and S41, calculating a keyword relevance score.
The manner of calculating the keyword relevance score in step S3 is the same as that in step S41. The traditional calculation process of TDF in excess formation by using the TF-IDF algorithm to calculate the correlation is too time-consuming, and the TF-IDF algorithm does not show the position information of words, so the invention improves the TF-IDF algorithm to calculate the keyword correlation score.
The method comprises the steps of setting given promoted article information to comprise three types of labels including product titles, category names and description information, calculating word frequencies of keywords in the product titles, the category names and the description information, constructing a judgment matrix by using an analytic hierarchy process to obtain criterion weights corresponding to the three types of labels, wherein the weights of the product titles, the category names and the description information calculated in one embodiment are [0.104729, 0.636986 and 0.258285 respectively ], multiplying the weights by the word frequencies of the keywords under the corresponding labels to obtain new word frequencies reflecting the importance of different labels, and then comprehensively calculating the relevance with indexes replacing IDFs.
Step S41 specifically includes the following sub-steps:
s411, performing word segmentation on the selected keywords to obtain a sub-keyword sequence.
For example, if the currently selected keyword is "new style one-piece dress", the sub-keywords after word segmentation are "new style" and "one-piece dress", and the two sub-keywords form a sub-keyword sequence in sequence.
S412, calculating the closeness score of the selected keyword according to the sub-keyword sequence and the given promoted item information.
Specifically, the position of the sub-keyword in the title or description of the given promotional item information is searched, the distance between the sub-keywords is calculated according to the word number of the sub-keyword after word segmentation in the given promotional item information, if the word number is adjacent, the compactness is full, if a certain score is deducted at intervals, for example, if the distance exceeds 20 words (about 10 words), the word number is regarded as not adjacent, and the score is 0. In addition, if the number of words of the sub-keyword after word segmentation does not exceed 20 words in the title or description, the relevance is considered to be too low, and the term is directly 0.
The closeness score ρ is calculated as:
wherein MaxForScore is a preset maximum scoring distance, max is the maximum distance value in the sub-keywords smaller than the maximum scoring distance, remove is the number of words in the sub-keywords that are not related to the given promotional item information or whose distance exceeds the maximum scoring distance, and remove Max is the upper limit of remove.
S413, calculating the relevance score of each sub keyword according to the closeness score of the selected keyword.
The formula for calculating the relevance score of each sub-keyword is as follows:
wherein, the first and the second end of the pipe are connected with each other,
j =1,2, \ 8230for the relevance score of the jth sub-keyword, t +1 is the total number of the sub-keywords of the selected keyword, and f is the total word frequency of the sub-keywords of the selected keyword in the given promotion item information; the total word frequency of the sub-keywords of the selected keyword in the given promotional item information is obtained by multiplying the word frequency of the corresponding keyword in the given promotional item information by the importance weight of the tag to which the corresponding keyword belongs.
The word frequency and the compactness range are both [0,1 ]]I.e. by
PVScore is the browsing volume score, i.e. the ratio of the browsing volume of the currently selected keyword to the average value of the browsing volume in the search ranking database, and if PV is greater than the average value, PVScore =1; CTRSCore is the click-through rate score, i.e., the click-through rate of the currently selected keywordSearching the ratio of the average value of the click rate in the ranking database, if CTR is more than the average value, CTRSCore =1, and the value range of the correlation score of the sub-keywords is
。
And S414, integrating the relevance scores of the sub-keywords, calculating the comprehensive score of the selected keyword, and taking the comprehensive score of the selected keyword as the keyword relevance score of the selected keyword.
Through the calculation in step S413, each sub-keyword after the word segmentation of the selected keyword has its own relevance score, and the present invention integrates the scores of all sub-keywords by a gradual increase method to obtain the score of the selected keyword before the word segmentation, and embodies the role of each specific sub-keyword score in the total score. The specific way of integrating the scores is as follows: the score of the first sub-keyword + the score of the first sub-keyword not obtained + the score of the second sub-keyword obtained in the ratio of the total score, for example, the score sequence corresponding to the sub-keyword sequence is (8, 3), the first score is 8+ (full score 10-score 8) ((second score 3/full score 10) = 8.6), and the longer score sequence is the same.
Specifically, let the sequence { x } 1 ,x 2 ,…,x t ,x t+1 And f, the correlation score sequence of the sub-keyword sequence is obtained, t +1 is the total number of the sub-keywords, and the correlation scores of the sub-keywords are integrated by adopting the following recursive formula:
wherein, { S 1 ,S 2 ,…,S t The integrated score sequence is used as the score sequence,
is the composite score of the selected keyword.
And S42, calculating the product relevance score of the selected keyword.
And searching all product related words according to given popularization article information to form a product related word set, searching related words of the currently selected keywords from the product related word set to form a keyword related word set, wherein the logic of searching the keyword related words is the same as that in the step S2. And judging whether the keyword related word set and the selected keyword have an inclusion relationship, and calculating the product relevance score of the selected keyword according to the keyword relevance score of the selected keyword.
If the current selected keyword exists in the keyword related word set, the product relevance score is equal to the keyword relevance score of the current selected keyword;
if the current selected keyword does not exist in the keyword related word set, the product relevance score is equal to the maximum keyword relevance score in the keyword related word set multiplied by 0.5.
And S43, calculating the creative relevance score of the selected keyword.
The method comprises the steps of segmenting given popularization article information, judging whether the selected keywords and sub-keywords segmented by the given popularization article information have inclusion or included relations, calculating keyword relevance scores of the sub-keywords with the inclusion or included relations, and calculating creative relevance scores according to the keyword relevance scores of the sub-keywords.
Specifically, the creative flow distribution mode can be divided into "preferred", "carousel", "intelligent" and "common", and one promoted item can have a plurality of creatives.
Splitting the given popularization article information, and judging the inclusion or included relationship between the selected keyword and the sub-keyword after splitting the popularization article information. For example, assuming that the given promotional item information is "summer new version korean dress", the currently selected keyword is "new version dress", the given promotional item information is split into "summer", "new version", "korean version", "one-piece dress", the selected keyword "new version dress" includes "new version", "one-piece dress", and the current creative score is equal to the largest keyword correlation score among the two words "new version", "one-piece dress";
if the inclusion or included logic is not satisfied, the current creative score is equal to the average of the keyword relevance scores of all the split words;
if the creative traffic distribution mode is "preferred" or "intelligent", the final creative relevance score is equal to the maximum of all creative relevance scores;
if the creative traffic distribution mode is 'carousel', the final creative relevance score is equal to the average value of all creative relevance scores;
if the creative traffic distribution style is "normal," then the final creative relevance score is equal to the first creative relevance score.
And S44, calculating the product quality score of the selected keyword.
And respectively inputting each selected keyword into a search ranking database for traversal, and querying products and related product attribute information corresponding to the selected keywords by segmenting the selected keywords, wherein the product attribute information comprises display amount, click rate, conversion rate, good rating number, total rating number, good rating rate and product rating. And respectively carrying out weighted summation on the product attribute information corresponding to each selected keyword to obtain the heat value of the selected keyword, and carrying out pre-estimation ranking on each selected keyword according to the heat value.
Extracting all average attribute values of all products under the category to which the current product belongs through a product related word set, and taking all the average attribute values as reference attribute values of corresponding attributes;
in the embodiment of the present invention, the attribute weight ratios of the display amount, the click amount, the conversion amount, the click rate, the conversion rate, the good score number, the total score number, the good score rate, and the product score are respectively 8.
Specifically, let the weight ratio of the attribute i be W i Where i =1,2, \ 8230, M, M is the total number of attributes,
the current product belongs toThe attribute value of the attribute i in the attribute information is R i The reference attribute value is R i_0 Then the score T of each attribute i i The calculation formula of (c) is:
and calculating the sum of the scores of the attributes as the quality score of the selected keyword.
Taking the calculation of the exhibited amount (i = 1) as an example, the average value of the exhibited amounts of all the products in the category to which the current product belongs is set to 5000, that is, the reference exhibited amount is R 1_0 =5000：
If the current product display amount is R 1 =10000, then R 1 ≥R 1_0 Showing amount score T 1 = W 1 =8；
If the current product display amount is R 1 =1000, then R 1 <R 1_0 Showing amount score T 1 = W 1 *R 1 /R i_0 =8*1000/5000=1.6。
And sequentially calculating each attribute score in the above manner, and adding the attribute scores to obtain the product quality score corresponding to the selected keyword.
And S5, weighting and summing the product relevance score, the creative relevance score and the product quality score to obtain the keyword quality score of each selected keyword.
The value ranges of the product correlation score, the creative correlation score, the product quality score and the keyword quality score are all [0,10].
And S6, sequencing all the selected keywords according to the quality scores of the keywords from high to low, and displaying sequencing results in a list form.
And sequencing all selected keywords selected by a target user according to the quality scores of the keywords from high to low, displaying sequencing results in a list form, and simultaneously displaying estimated bids, display amounts, click volumes, click rates, volume of deals, rate of deals and other data related to the selected keywords.
And S7, selecting the selected keywords with the keyword quality scores larger than a first preset threshold value and the estimated ranks smaller than a second preset threshold value from the sorting results as the optimal promotion keywords to be recommended to the target user.
According to the method and the device, the keyword quality score and the estimated ranking of the selected keyword are calculated according to the search heat corresponding to the selected keyword and the calculated corresponding product correlation score, the creative correlation score and the product quality score, the promotion degree of the keyword can be visually simulated through numerical sorting such as the keyword quality score and the estimated ranking, the keyword promotion degree can be helped to be selected by a target user, for example, the selected keyword with certain promotion value, the keyword with the keyword quality score larger than 8 and the estimated ranking smaller than 1000, is selected to serve as the promotion keyword of a promoted article, and the user is prevented from blindly screening the promotion keyword.
Corresponding to the embodiment of the method, the invention also provides a system for simulating and screening the promoted keywords, which comprises the following steps:
a database construction module: the system comprises a search keyword set, a search ranking database and a database server, wherein the search keyword set is used for acquiring a search keyword set of a user group through big data and establishing the search ranking database according to the search keyword set;
a keyword screening module: the system is used for recommending a candidate keyword set for a user according to given promoted item information; obtaining a plurality of selected keywords screened from a candidate keyword set by a target user;
the scoring and ranking module: the system is used for calculating the product relevance score, the creative relevance score and the product quality score of the selected keywords; calculating the keyword quality score of each selected keyword based on the product correlation score, the creative correlation score and the product quality score in a weighted summation manner; and sequencing all the selected keywords from high to low according to the quality scores of the keywords, and displaying the sequencing result in a list form.
The above system embodiments and method embodiments are in one-to-one correspondence, and please refer to the method embodiment for the brief description of the system embodiments.
The invention also discloses an electronic device, comprising: at least one processor, at least one memory, a communication interface, and a bus; the processor, the memory and the communication interface complete mutual communication through the bus; the memory stores program instructions executable by the processor which invokes the method of the invention as described above.
The invention also discloses a computer readable storage medium which stores computer instructions, and the computer instructions enable the computer to realize all or part of the steps of the method of the embodiment of the invention. The storage medium includes: various media capable of storing program codes, such as a U disk, a removable hard disk, a read-only memory ROM, a random access memory RAM, a magnetic disk, or an optical disk.
The above-described system embodiments are merely illustrative, wherein the units described as separate parts may or may not be physically separate, and the parts shown as units may or may not be physical units, i.e. may be distributed over a plurality of network units. Those skilled in the art can select some or all of the modules according to actual needs to achieve the purpose of the solution of the present embodiment without creative efforts.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and should not be taken as limiting the scope of the present invention, which is intended to cover any modifications, equivalents, improvements, etc. within the spirit and scope of the present invention.
Claims (9)
1. A method for simulating and screening promotion keywords is characterized by comprising the following steps:
acquiring a search keyword set of a user group through big data;
constructing a search ranking database according to the search keyword set;
recommending a candidate keyword set for a target user according to given promoted article information;
obtaining a plurality of selected keywords screened from the candidate keyword set by the target user, and calculating keyword correlation scores of the selected keywords; the calculating the keyword relevance score of the selected keyword specifically includes:
performing word segmentation on the selected keyword to obtain a sub-keyword sequence;
calculating the closeness score of the selected keyword according to the sub-keyword sequence and the given promoted article information;
calculating the relevance score of each sub keyword according to the closeness score of the selected keyword;
integrating the relevance scores of all the sub-keywords, calculating the comprehensive score of the selected keyword, and taking the comprehensive score of the selected keyword as the keyword relevance score of the selected keyword;
calculating the product relevance score, the creative relevance score and the product quality score of the selected keyword;
weighting and summing the product relevance score, the creative relevance score and the product quality score to obtain the keyword quality score of each selected keyword;
and sequencing all the selected keywords from high to low according to the quality scores of the keywords, and displaying the sequencing result in a list form.
2. The method for simulating and screening promotional keywords according to claim 1, wherein the constructing of the search ranking database according to the set of search keywords comprises:
acquiring search keywords of an nth user in a user group, and extracting core keywords of the search keywords, wherein N =1,2, \ 8230;
extracting product related words and keyword related words of the search keywords according to the core keywords;
forming the core keywords, the product related words and the keyword related words into nth search words;
obtaining the nth search result information according to the probability of each product in the search result of the nth search word;
repeating the steps until the Nth search result information is obtained;
and collecting and summarizing all search result information to form a search ranking database.
3. The method for simulating and screening popularization keywords according to claim 2, wherein the search results corresponding to each search keyword in the search ranking database carry a plurality of attribute information, including presentation amount, click rate, conversion rate, goodness score, total goodness score, goodness score and product score.
4. The method for simulating and screening promotional keywords according to claim 1 wherein the calculating of the closeness score of the selected keyword based on the sequence of sub-keywords and given promotional item information specifically comprises:
searching the position of the sub-keyword sequence in the given popularization article information, calculating the distance between the sub-keywords according to the word numbers of the sub-keywords after word segmentation in the given popularization article information, wherein the calculation formula of the compactness score rho is as follows:
wherein MaxForScore is a preset maximum scoring distance, max is the maximum distance value in the sub-keywords smaller than the maximum scoring distance, remove is the number of words in the sub-keywords that are not related to the given promotional item information or whose distance exceeds the maximum score distance, and removeMax is the upper limit of remove.
5. The method for simulating and screening promotional keywords according to claim 4, wherein the calculating the relevance score of each sub-keyword based on the closeness score of the selected keyword comprises:
wherein the content of the first and second substances,
the correlation score of the jth sub-keyword is j =1,2, \8230;, t +1 is selectedF is the total word frequency of the sub-keywords of the selected keyword in the given promotion article information, and rho is the compactness score; PVScore is a browsing volume score, i.e., a ratio of the browsing volume of the currently selected keyword to an average value of the browsing volume in the search ranking database, CTRScore is a click rate score, i.e., a ratio of the click rate of the currently selected keyword to an average value of the click rate in the search ranking database,
。
6. the method for simulating and screening promotional keywords according to claim 5, wherein the integrating the relevance scores of each sub-keyword and the calculating the composite score of the selected keyword specifically comprises:
let sequence { x 1 ,x 2 ,…,x t ,x t+1 And f, the correlation score sequence of the sub-keyword sequence is obtained, t +1 is the total number of the sub-keywords, and the correlation scores of the sub-keywords are integrated by adopting the following recursive formula:
wherein, { S 1 ,S 2 ,…,S t The integrated score sequence is used as the score sequence,
a composite score for the selected keyword.
7. The method of claim 3, wherein the calculating the product relevance score, the creative relevance score, and the product quality score for the selected keyword specifically comprises:
searching all product related words in a search ranking database according to given promoted article information to form a product related word set, searching related words of the currently selected keywords from the product related word set to form a keyword related word set; judging whether the keyword related word set and the selected keyword have an inclusion relationship, and calculating the product relevance score of the selected keyword according to the keyword relevance score of the selected keyword;
segmenting the given popularization article information, judging whether the selected keyword and the sub-keyword after segmentation of the given popularization article information have an inclusion or included relationship, calculating a keyword correlation score of the sub-keyword having the inclusion or included relationship, and calculating an creative correlation score according to the keyword correlation score of the sub-keyword;
and comparing the product attribute information corresponding to the selected keyword with each average attribute value of all products under the category to which the current product belongs, calculating the score of each attribute based on the set product attribute weight ratio, and taking the sum of the scores of the attributes as the product quality score of the selected keyword.
8. The method for simulating and screening promotional keywords according to claim 7, wherein the comparing the product attribute information corresponding to the selected keyword with the respective average attribute values of all products under the category to which the current product belongs, and the calculating the score of each attribute based on the set product attribute weight ratio specifically comprises:
let the weight ratio of the attribute i be W i ，
The attribute value of the attribute i in the current product attribute information is R i The average attribute value of each of all products in the category to which the current product belongs is R i_0 Then the score T of each attribute i i The calculation formula of (c) is:
where i =1,2, \ 8230, and M, M is the total number of attributes.
9. A promotion keyword simulation screening system, the system comprising:
a database construction module: the system comprises a search keyword set, a search ranking database and a database server, wherein the search keyword set is used for acquiring a search keyword set of a user group through big data and establishing the search ranking database according to the search keyword set;
a keyword screening module: the system is used for recommending a candidate keyword set for a target user according to given promoted item information; obtaining a plurality of selected keywords screened from a candidate keyword set by a target user;
the scoring and ranking module: for calculating a keyword relevance score for the selected keyword; calculating the product relevance score, the creative relevance score and the product quality score of the selected keyword; weighting and summing the product relevance score, the creative relevance score and the product quality score to obtain the keyword quality score of each selected keyword; sorting all the selected keywords in the sequence from high to low according to the quality scores of the keywords, and displaying the sorting results in a list form; the calculating the keyword relevance score of the selected keyword specifically includes: performing word segmentation on the selected keyword to obtain a sub-keyword sequence; calculating the closeness score of the selected keyword according to the sub-keyword sequence and the given promoted article information; calculating the relevance score of each sub keyword according to the closeness score of the selected keyword; and integrating the relevance scores of the sub-keywords, calculating the comprehensive score of the selected keyword, and taking the comprehensive score of the selected keyword as the keyword relevance score of the selected keyword.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211268303.2A CN115329760B (en) | 2022-10-17 | 2022-10-17 | Method and system for simulating and screening promotion keywords |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211268303.2A CN115329760B (en) | 2022-10-17 | 2022-10-17 | Method and system for simulating and screening promotion keywords |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115329760A true CN115329760A (en) | 2022-11-11 |
| CN115329760B CN115329760B (en) | 2023-02-14 |
Family
ID=83915249
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211268303.2A Active CN115329760B (en) | 2022-10-17 | 2022-10-17 | Method and system for simulating and screening promotion keywords |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115329760B (en) |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103226618A (en) * | 2013-05-21 | 2013-07-31 | 焦点科技股份有限公司 | Related word extracting method and system based on data market mining |
| CN106484698A (en) * | 2015-08-25 | 2017-03-08 | 北京奇虎科技有限公司 | A kind of method for pushing of search keyword and device |
| CN114925261A (en) * | 2022-05-20 | 2022-08-19 | 深圳前海微众银行股份有限公司 | Keyword determination method, apparatus, device, storage medium and program product |
-
2022
- 2022-10-17 CN CN202211268303.2A patent/CN115329760B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103226618A (en) * | 2013-05-21 | 2013-07-31 | 焦点科技股份有限公司 | Related word extracting method and system based on data market mining |
| CN106484698A (en) * | 2015-08-25 | 2017-03-08 | 北京奇虎科技有限公司 | A kind of method for pushing of search keyword and device |
| CN114925261A (en) * | 2022-05-20 | 2022-08-19 | 深圳前海微众银行股份有限公司 | Keyword determination method, apparatus, device, storage medium and program product |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115329760B (en) | 2023-02-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102902691B (en) | Recommend method and system | |
| CN104866474B (en) | Individuation data searching method and device | |
| Lee et al. | Mining perceptual maps from consumer reviews | |
| CN108628833B (en) | Method and device for determining summary of original content and method and device for recommending original content | |
| CN103593425B (en) | Preference-based intelligent retrieval method and system | |
| CN103246980B (en) | Information output method and server | |
| CN107424043A (en) | A kind of Products Show method and device, electronic equipment | |
| US20120215776A1 (en) | Query identification and association | |
| CN108256119A (en) | A kind of construction method of resource recommendation model and the resource recommendation method based on the model | |
| CN106776860A (en) | One kind search abstraction generating method and device | |
| CN110532351B (en) | Recommendation word display method, device and equipment and computer readable storage medium | |
| JP2004164578A (en) | Method and apparatus for categorizing and presenting document of distributed database | |
| KR20050049750A (en) | On-line advertising system and method | |
| WO2007070199A1 (en) | Advertising keyword cross-selling | |
| TW201207651A (en) | Operationalizing search engine optimization | |
| US20100318427A1 (en) | Enhancing database management by search, personal search, advertising, and databases analysis efficiently using core-set implementations | |
| CN107563867A (en) | A kind of commending system cold start-up method based on multi-arm fruit machine confidence upper limit | |
| CN104391883A (en) | Online advertisement audience sorting method based on transfer learning | |
| Krestel et al. | Diversifying customer review rankings | |
| Yankov et al. | Evaluation of explore-exploit policies in multi-result ranking systems | |
| CN106339898A (en) | Product innovation method based on internet big data | |
| Lin et al. | A consumer review-driven recommender service for web e-commerce | |
| CN112862567A (en) | Exhibit recommendation method and system for online exhibition | |
| US20190065611A1 (en) | Search method and apparatus | |
| KR20050050016A (en) | On-line advertising system and method |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CP03 | Change of name, title or address | ||
| CP03 | Change of name, title or address |
Address after: No.568, longmian Avenue, Jiangning District, Nanjing, Jiangsu 210000 Patentee after: Nanjing Zhongjiao Changxiang Internet Technology Co.,Ltd. Country or region after: China Patentee after: Zhongjiao Changxiang Technology Co.,Ltd. Address before: No.568, longmian Avenue, Jiangning District, Nanjing, Jiangsu 210000 Patentee before: Nanjing Zhongjiao Changxiang Internet Technology Co.,Ltd. Country or region before: China Patentee before: Chinese Education Changxiang (Beijing) Technology Co.,Ltd. |