CN105550901A - Few user evaluations based goods popularity and preference combined prediction system - Google Patents
Few user evaluations based goods popularity and preference combined prediction system Download PDFInfo
- Publication number
- CN105550901A CN105550901A CN201510968255.1A CN201510968255A CN105550901A CN 105550901 A CN105550901 A CN 105550901A CN 201510968255 A CN201510968255 A CN 201510968255A CN 105550901 A CN105550901 A CN 105550901A
- Authority
- CN
- China
- Prior art keywords
- user
- article
- popularity
- prediction
- small amount
- 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.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- 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
Abstract
The invention discloses a few user evaluations based goods popularity and preference combined prediction system. The system comprises a data preparation module, a feature modeling module and a prediction module. The data preparation module preprocesses original evaluation data of users to generate a user-goods score matrix and a goods popularity vector. The feature modeling module generates a preference feature matrix of users to different goods, a goods attribute feature matrix, a goods attribute matrix and a popularity regression coefficient vector. The prediction module maps the preference feature matrix and new goods evaluations of a few key users into an attribute feature vector of new goods, and uses the attribute feature vector of the new goods to be multiplied by a linear prediction coefficient to obtain a future popularity prediction result. According to the system, new goods popularity and new goods preference in all users are subjected to combined prediction only by adopting the evaluations of the few ''key'' users with remarkable distinction degrees, so that the running speed is increased, the time cost is reduced, and the accuracy of prediction is ensured.
Description
Technical field
The present invention relates to a kind of field of computer technology, specifically, what relate to is a kind ofly utilize computer calculate when only obtaining and utilize minute quantity Given information to the popularity of article and user preferences degree associated prediction system.
Background technology
In the epoch that ecommerce is constantly risen, user preferences degree prognoses system has become the important component part of stimulating customer consumption.So-called preference degree prediction, refer to that businessman is according to the history shopping list of client that have purchased a part of commodity, guess that other do not buy fancy grade or the taste of commodity or new commodity in website for he, and commercial product recommending higher for prediction scoring to client, it is bought with sales promotion.The personal marketing of this accurate formula has become the only magic weapon of many Internet firms, e-commerce website, advertising provider, plays very important role in internet economy.
Except the prediction of user individual hobby, in goods marketing, advertising company and shopping website also need this important indicator of potential popularity checking and rating commodity, it reflects the popular overall acceptance level for commodity.Cite a plain example, if a commodity only have higher fancy grade in client very among a small circle, so it just can regard a minority's commodity as, is tilted over many advertising promotion resources can not brings larger interests to it.Compare, if other a commodity can all obtain good preference degree in fairly large client, the marketing so for him may have an economic benefit more.Potential popularity mentioned here, be not limited to the popularity value of article in set of system or a website in essence, it can be weighed with consumers' acceptable degree in larger scope.
Comprehensive above 2 consideration, a successful commodity projection system not only needs to recognize to be evaluated for the user habit preference of each sole user, prediction user, also needs the popularity situation of understanding article in total system.It should be noted that, this prognoses system is not limited to Internet firm, and in any entity commercial economy body, the sale making precision according to client's preference degree is also the important step of stimulating customer consumption.Be more importantly, the prediction of the prediction of popularity and user's scoring can be even the combination of internet economy and real economy, such as film achieves very good achievement in box office at the cinema, so it just has very high popularity, and an online film-on-demand website just can utilize these popularity information, and like according to the personalization of oneself website user, comprehensively choose optimum film and carry out ad promotions.The predictive mode of this amalgamation has very important economic worth.
Due to the existing evaluation information of new commodity and attributive character less, this openness feature that result in the similarity trend zero of user-article, cannot obtain the model parameter of commodity accurately.Therefore existing prognoses system needs are by method acquisition user groups in a big way such as collection, market studys to the evaluation of new commodity, can calculate user-article similarity preferably, thus carry out follow-up prediction work.And this pattern needs very large economy and time cost, data processing complex, computing efficiency is low.
Because new commodity emerges in an endless stream, what cause because the evaluation of new commodity is very few in the prediction of current article popularity and user preferences degree prognoses system cannot its popularity of Accurate Prediction and the preference degree in whole user group, and a kind of new technological means of eager needs deals with the Popularity prediction of commodity and the problem of user preferences degree prediction.
Summary of the invention
The object of the invention is to for the deficiencies in the prior art, propose a kind of article popularity of evaluating based on a small amount of user and preference degree associated prediction system, this system utilizes computer technology, based on a small amount of key user chosen in advance, the evaluation of new commodity is predicted, reduce data processing complex, improve computing efficiency, a large amount of manpower and materials can be saved, and accuracy is high.
The present invention is achieved by the following technical solutions, and described prognoses system comprises:
Data preparation module: by the iotave evaluation data prediction of user, generates user-article rating matrix, article popularity vector;
Feature modeling module: according to the user-article rating matrix of data preparation module, article popularity vector, adopts functional matrix disassembling method, linear regression method to generate the regression coefficient vector of user to the hobby eigenmatrix of different article, goods attribute eigenmatrix, goods attribute matrix and popularity;
Prediction module: only use a small amount of key user to the evaluation of article, the proper vector of estimation goods attribute, thus the following popularity of prediction prediction article, prediction total user are to the fancy grade of these article; Described prediction module the hobby eigenmatrix of a small amount of key user, a small amount of key user to the evaluation of new article, be mapped as the attribute feature vector of this new article, then use the attribute feature vector of new article and total user to like eigenmatrix to be multiplied and to predict total user fancy grade, use the attribute feature vector of new article to be multiplied with linear predictor coefficient and obtain following Popularity prediction result.
Preferably, described data preparation module, feature modeling module, the article wherein related to, refer to the commodity be now present in this system;
Described prediction module, the article wherein related to comprise the commodity be now present in this system, to be not present in current system, the new article that can be injected towards current system.
Preferably, described data preparation module, feature modeling module, when pre-service wherein and founding mathematical models, by user to the figure of merit of certain article, be normalized to the element (being then considered as sky element if there is not evaluation) intersected in user-article rating matrix; Various dimensions goods attribute feature is mapped as the popularity vector of one dimension by linear predictor coefficient, can also pop degree be vectorial carries out nonlinear transformation process and make it be within certain variation range.
Preferably, described prediction module, comprise and choose a small amount of key user's submodule, the user preferences eigenmatrix that this submodule obtains according to described data preparation module, feature modeling module, goods attribute eigenmatrix, regression coefficient vector, select the user of a small amount of optimum (namely make prediction error rate minimum) as key user from existing system in whole user.Only use these a small amount of key users to the evaluation of new article subsequently, predict the popularity of these article and the fancy grade in total user, thus time cost required when greatly reducing associated prediction, human cost etc., and keep the accuracy of prediction.
Preferably, described chooses a small amount of key user's submodule, the problem that optimization chooses a small amount of key user can be converted into the problem building and choose optimum branching node in three-dimensional decision-tree model.This decision-tree model, construction method is:
1) in decision tree, each leaf node comprises the similar article of one group of attributive character.Each branch node of decision tree every layer comprises a user.According to the evaluation (good, bad, uncertain) of each user to article in child node, all can divide these article three subsets divided into without occuring simultaneously, the three son nodes as current decision tree node: left sibling L (p), interior joint U (p), right node D (p).In whole user, the article popularity predicated error in three subsets after a user p must be had can to make graduation and user preferences degree error sum are minimum, this user p are designated as the node of this branch of decision tree, and are considered as a key user;
2) identical recursive subdivision operation is all adopted to each child of decision tree every layer, the decision tree that maximum level is P layer can be built.When needing the attributive character obtaining article to be predicted, according to P key user to the evaluation of these article, most suitable decision path can be selected, and arrives among corresponding leaf node, and then obtain this goods attribute feature, for use in prediction.
Present system achieves following characteristics:
1) according to user known in system to the evaluation of existing commodity and hobby, generate and can distinguish the mathematical model of different user to different commodity preference degree, and generate and can distinguish the mathematical model of different attribute feature between commodity;
2) this system can choose one group of a small amount of key user in advance.For new commodity to be predicted, this group user only can be utilized the evaluation of this new commodity, estimate the attributive character model of these commodity, and predict that in the popularity of these commodity, system, all users are to the fancy grade of this new commodity successively.
3) this system binding function matrix decomposition and linear prediction method, can most optimally choose above-mentioned key user group, therefore can ensure Popularity prediction accuracy and recommend accuracy.
The system predicted just can be carried out compared to the existing collection more users evaluation information that needs, the present invention has following beneficial effect: the present invention builds the decision tree that optimization chooses a small amount of crucial evaluation and test user, and utilize a small amount of but there is the evaluation of user to new commodity of discrimination, predict popularity and the total user fancy grade of this new commodity.The technical method that native system comprises not only significantly reduces data volume process, improves travelling speed, reduces systematic cost, more pop degree and user preferences can carry out associated prediction simultaneously simultaneously, and can not reduce predictablity rate.
Accompanying drawing explanation
By reading the detailed description done non-limiting example with reference to the following drawings, other features, objects and advantages of the present invention will become more obvious:
Fig. 1 is system chart in one embodiment of the invention;
Fig. 2 is the workflow diagram of system in one embodiment of the invention;
Fig. 3 is the process flow diagram of the system of selection of optimization key user in one embodiment of the invention.
Embodiment
Below in conjunction with specific embodiment, the present invention is described in detail.Following examples will contribute to those skilled in the art and understand the present invention further, but not limit the present invention in any form.It should be pointed out that to those skilled in the art, without departing from the inventive concept of the premise, some distortion and improvement can also be made.These all belong to protection scope of the present invention.
As shown in Figure 1, a kind of article popularity of evaluating based on a small amount of key user and user preferences degree associated prediction system, comprising:
Data preparation module: for by iotave evaluation data prediction, generate user-article rating matrix, article popularity vector;
Feature modeling module: according to the user-article rating matrix of data preparation module, article popularity vector, adopts functional matrix disassembling method, linear regression method to generate the regression coefficient vector of user to the hobby eigenmatrix of different article, goods attribute eigenmatrix, goods attribute matrix and popularity;
Prediction module: only use a small amount of key user to the evaluation of article, the proper vector of estimation goods attribute, thus the following popularity of prediction prediction article, prediction total user are to the fancy grade of these article.
Described data preparation module, feature modeling module, the article wherein related to, refer to the commodity be now present in this system;
Described prediction module, the article wherein related to comprise the commodity be now present in this system, to be not present in current system, the new article that can be injected towards current system.
Article popularity of the present invention is not limited to the sales volume etc. of commodity in triangular web, can refer to sales volume in wider multiple system or overall market or weigh other numerical value etc. of popularity degree, trend.
As shown in Figure 2, the course of work of said system prediction is:
1: the data preparation stage of data preparation module.
Each user in acquisition system for the evaluation of article, the popularity information of each commodity in acquisition system.After raw data pre-service, generate user-article rating matrix, article popularity vector.
The evaluation of user and article popularity can be obtained by artificial or machine mode.And weigh the fancy grade of user to commodity by the evaluation of user being mapped as the scoring adopting 1-5 to divide, and produce user-article scoring sparse matrix R (be wherein set to sky without the element of marking or value is zero) thus, but be not limited to this mode.
The popularity of commodity can adopt the modes such as third party's index, sales volume to define, and through data value maps to the numerical value changed in certain fixed range.And remember that the popularity variable of all commodity is b.Obtain manner, the variation of popularity are not limited to aforesaid way.
2: the user preferences feature of feature modeling module, article popularity feature modeling stage.
According to Evaluations matrix, the article popularity vector of user known in system to existing article, generate user to hobby distance matrix, the article characteristics distance matrix of different article by functional matrix disassembling method, so as to set up can distinguish different user, without product features mathematical model.
In the user-article Evaluations matrix R supposing to have N number of user and M commodity to form, represent the evaluation of user to commodity, its neutralization is considered as the model parameter of user, commodity respectively, is K dimensional vector.Whole user, goods model parameter are designated as U, V.The scoring of product representation user in the model parameter that so can use the model parameter of user, commodity, and the unknown element in R also can by obtaining.Meanwhile, object model parameter U can also be article popularity vector by popularity regression model coefficient mapping, is also.
Suppose the scoring set of user known in representing matrix R to commodity j, so wanting to obtain whole unknown element (namely user marks), popularity regression model coefficient in R needs to solve following optimization problem to reach Popularity prediction, score in predicting result the most accurately:
This optimizing process will be solved by the optimization of successive ignition, i.e. following 3 steps:
Known, upgrade; Known, upgrade U; Upgrade.
Solve in the operation of above-mentioned optimization problem actual, native system can the L1 norm of pop degree regression model coefficient retrain, to avoid the excessive generation over-fitting of model coefficient domain of walker further.The span of this constraint can regulate according to the scale N of real data, M and model parameter dimension K.
3: the crucial guide user of prediction module chooses the stage:
This stage need combine consider user area calibration, Popularity prediction accuracy, recommendation results accuracy after from the crucial guide user selected existing system, therefore need the rating matrix R according to upper step, article popularity vector, decompose and generate most suitable U, V matrix numerical value.
Native system is designated as T the mapping function that the evaluation of crucial guide user to commodity is mapped as the model parameter of commodity.So in order to make user model parameter (hobby distance matrix U) produce maximum discrimination, the problem that optimization chooses guide user can be converted into the problem (as shown in Figure 3) building and choose optimum branching node in three-dimensional decision-tree model.
Branch node in decision tree is considered as a guide user p.All commodity are considered as leaf node.According to the evaluation (good, bad, uncertain) of guide user to commodity, all commodity are divided into three parts, and become three son nodes of current decision tree node: left sibling L (p), interior joint U (p), right node D (p).In whole user, the predicated error that must have a user that this graduation can be made to produce is minimum, regards it as guide user p.Prediction error functions can calculate with following formula:
Identical recursive operation is adopted to each child, thus builds the decision tree that complete maximum level is P layer.
4: the output stage that the prediction of the article popularity of prediction module, article are recommended.
For the article m to be predicted that each is new, first native system obtains an above-mentioned P guide user to the evaluation of these article to be predicted, and according to the user characteristics model of this P guide user, generates the proper vector of this new article.The prediction obtaining total user by the inner product of total user model parameter and this article characteristics vector is subsequently marked, and recommends targeted customer according to this scoring; The predict popularity of these article is obtained according to the product of popularity regression model coefficient and new article proper vector.In general, the article of high popularity can obtain liking of large-scale consumer, but the customer group degree of overlapping of this kind of commodity is higher; Difference " minority " article of low popularity then can obtain higher preference degree in different microcommunities.Two prediction index that can export according to the motor-driven reference native system of actual conditions, to select suitable potential user group.
The present invention proposes and only adopt the evaluation of " key " user on a small quantity with remarkable discrimination preference degree in total user carries out associated prediction to new article popularity and new article, this associated prediction system have employed such as functional matrix decomposition, the technical skills such as linear regression build and predicated error can be made to fall minimum optimal user trade-off decision tree-model, and be applied to associated prediction, which system reduces the complexity of data processing, improve travelling speed, greatly can reduce labour cost required for market study after new commodity puts goods on the market, reduce time cost, but do not reduce the accuracy of prediction simultaneously.
Above specific embodiments of the invention are described.It is to be appreciated that the present invention is not limited to above-mentioned particular implementation, those skilled in the art can make various distortion or amendment within the scope of the claims, and this does not affect flesh and blood of the present invention.
Claims (6)
1. the article popularity evaluated based on a small amount of user and a preference degree associated prediction system, is characterized in that comprising:
Data preparation module: by the iotave evaluation data prediction of user, generates user-article rating matrix, article popularity vector;
Feature modeling module: according to the user-article rating matrix of data preparation module, article popularity vector, adopts functional matrix disassembling method, linear regression method to generate the regression coefficient vector of user to the hobby eigenmatrix of different article, goods attribute eigenmatrix, goods attribute matrix and popularity;
Prediction module: only use a small amount of key user to the evaluation of article, the proper vector of estimation goods attribute, thus the following popularity of prediction prediction article, prediction total user are to the fancy grade of these article;
Described prediction module the hobby eigenmatrix of a small amount of key user, a small amount of key user to the evaluation of new article, be mapped as the attribute feature vector of this new article, then use the attribute feature vector of new article and total user to like eigenmatrix to be multiplied and to predict total user fancy grade, use the attribute feature vector of new article to be multiplied with linear predictor coefficient and obtain following Popularity prediction result.
2. the article popularity evaluated based on a small amount of user according to claim 1 and preference degree associated prediction system, is characterized in that:
Described data preparation module, feature modeling module, the article wherein related to, refer to the commodity be now present in this system;
Described prediction module, the article wherein related to comprise the commodity be now present in this system, to be not present in current system, the new article that can be injected towards current system.
3. the article popularity evaluated based on a small amount of user according to claim 1 and preference degree associated prediction system, it is characterized in that: described data preparation module, feature modeling module, when pre-service wherein and founding mathematical models, by user to the figure of merit of certain article, be normalized to the element intersected in user-article rating matrix; Various dimensions goods attribute feature is mapped as the popularity vector of one dimension by linear predictor coefficient, and pop degree vector carries out nonlinear transformation process and makes it be within certain variation range.
4. the article popularity evaluated based on a small amount of user according to any one of claim 1-3 and preference degree associated prediction system, it is characterized in that: described prediction module, comprise and choose a small amount of key user's submodule, this submodule is according to described data preparation module, the user preferences eigenmatrix that feature modeling module obtains, goods attribute eigenmatrix, regression coefficient vector, a small amount of optimum user is selected as key user from the whole user of existing system, only use these a small amount of key users to the evaluation of new article subsequently, predict the popularity of these article and the fancy grade in total user.
5. the article popularity evaluated based on a small amount of user according to claim 4 and preference degree associated prediction system, it is characterized in that: described chooses a small amount of key user's submodule, problem optimization being chosen a small amount of key user is converted into the problem building and choose optimum branching node in three-dimensional decision-tree model.
6. the article popularity evaluated based on a small amount of user according to claim 5 and preference degree associated prediction system, it is characterized in that: described decision-tree model, construction method is:
1) in decision tree, each leaf node comprises the similar article of one group of attributive character, each branch node of decision tree every layer comprises a user, according to the evaluation of each user to article in child node, equal can these article divide divide into without occur simultaneously three subsets, the three son nodes as current decision tree node: left sibling L (p), interior joint U (p), right node D (p); In whole user, the article popularity predicated error in three subsets after a user p must be had to make graduation and user preferences degree error sum are minimum, this user p are designated as the node of this branch of decision tree, and are considered as a key user;
2) identical recursive subdivision operation is all adopted to each child of decision tree every layer, build the decision tree that maximum level is P layer; When needing the attributive character obtaining article to be predicted, according to P key user to the evaluation of these article, trade-off decision path, and arrive among corresponding leaf node, and then obtain this goods attribute feature, for use in prediction.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510968255.1A CN105550901A (en) | 2015-12-21 | 2015-12-21 | Few user evaluations based goods popularity and preference combined prediction system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510968255.1A CN105550901A (en) | 2015-12-21 | 2015-12-21 | Few user evaluations based goods popularity and preference combined prediction system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN105550901A true CN105550901A (en) | 2016-05-04 |
Family
ID=55830080
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510968255.1A Pending CN105550901A (en) | 2015-12-21 | 2015-12-21 | Few user evaluations based goods popularity and preference combined prediction system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105550901A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108596695A (en) * | 2018-05-15 | 2018-09-28 | 口口相传(北京)网络技术有限公司 | Entity method for pushing and system |
| CN108846699A (en) * | 2018-06-14 | 2018-11-20 | 殷运林 | Demand matching process and device |
| CN109933749A (en) * | 2017-12-19 | 2019-06-25 | 北京京东尚科信息技术有限公司 | Method and apparatus for generating information |
| CN109977393A (en) * | 2017-12-28 | 2019-07-05 | 中国科学院计算技术研究所 | A kind of popular news prediction technique and system based on content controversial |
| CN111652674A (en) * | 2020-05-15 | 2020-09-11 | 拉扎斯网络科技（上海）有限公司 | Resource recommendation method and device |
-
2015
- 2015-12-21 CN CN201510968255.1A patent/CN105550901A/en active Pending
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109933749A (en) * | 2017-12-19 | 2019-06-25 | 北京京东尚科信息技术有限公司 | Method and apparatus for generating information |
| CN109933749B (en) * | 2017-12-19 | 2024-03-05 | 北京京东尚科信息技术有限公司 | Method and device for generating information |
| CN109977393A (en) * | 2017-12-28 | 2019-07-05 | 中国科学院计算技术研究所 | A kind of popular news prediction technique and system based on content controversial |
| CN108596695A (en) * | 2018-05-15 | 2018-09-28 | 口口相传(北京)网络技术有限公司 | Entity method for pushing and system |
| CN108596695B (en) * | 2018-05-15 | 2021-04-27 | 口口相传(北京)网络技术有限公司 | Entity pushing method and system |
| CN108846699A (en) * | 2018-06-14 | 2018-11-20 | 殷运林 | Demand matching process and device |
| CN111652674A (en) * | 2020-05-15 | 2020-09-11 | 拉扎斯网络科技（上海）有限公司 | Resource recommendation method and device |
| CN111652674B (en) * | 2020-05-15 | 2023-09-19 | 拉扎斯网络科技（上海）有限公司 | Resource recommendation method and device |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104391849B (en) | Incorporate the collaborative filtering recommending method of time contextual information | |
| CN105550901A (en) | Few user evaluations based goods popularity and preference combined prediction system | |
| CN103353872B (en) | A kind of teaching resource personalized recommendation method based on neutral net | |
| Ayub et al. | A Jaccard base similarity measure to improve performance of CF based recommender systems | |
| CN103632290B (en) | A kind of based on the mixing recommendation method recommending probability fusion | |
| CN108416625A (en) | The recommendation method and apparatus of marketing product | |
| CN106296242A (en) | A kind of generation method of commercial product recommending list in ecommerce and the system of generation | |
| Lai et al. | Novel personal and group-based trust models in collaborative filtering for document recommendation | |
| KR101963817B1 (en) | Apparatus and method for generating prediction information based on a keyword search volume | |
| CN104239338A (en) | Information recommendation method and information recommendation device | |
| CN109711925A (en) | Cross-domain recommending data processing method, cross-domain recommender system with multiple auxiliary domains | |
| CN103824213A (en) | Individualized recommendation method based on user preferences and commodity properties | |
| CN103150667B (en) | A kind of personalized recommendation method based on body construction | |
| CN106708868B (en) | Internet data analysis method and system | |
| CN105005701A (en) | Personalized recommendation method based on attributes and scores | |
| CN104657336A (en) | Personal recommendation algorithm based on semi-cosine function | |
| CN108198041A (en) | A kind of Method of Commodity Recommendation based on time attenuator, terminal and medium | |
| CN104615721B (en) | For the method and system based on return of goods related information Recommendations | |
| CN107679103A (en) | For entity attributes analysis method and system | |
| CN111340522B (en) | Resource recommendation method, device, server and storage medium | |
| CN107767152A (en) | Product purchase intention analysis method and server | |
| Wang et al. | Random forest analysis of factors affecting urban carbon emissions in cities within the Yangtze River Economic Belt | |
| CN114510653A (en) | Social group recommendation method, system, device and storage medium | |
| Zhang et al. | Application of sample balance-based multi-perspective feature ensemble learning for prediction of user purchasing behaviors on mobile wireless network platforms | |
| Hong et al. | Probabilistic reliable linguistic term sets applied to investment project selection with the gained and lost dominance score method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20160504 |
|
| RJ01 | Rejection of invention patent application after publication |