CN105678567A - Accurate prediction system based on big data deep learning - Google Patents

Abstract

The invention relates to an accurate prediction system based on big data deep learning. The accurate prediction system includes the following four parts: an information extraction system, a training prediction model system, a current user input information system and a predictor system. The accurate prediction system based on big data deep learning can realize accurate analysis of big data by means of the optimized deep learning technology; for a customer, the accurate prediction system can quickly search the required goods and reduce the candidate items, and can timely discover the new trend; and for a merchant, the accurate prediction system can perform promotion according to the interest of the customer, is individualized, can increase the customer loyalty and the sales amount, and can know about the customer and the market more deeply.

Description

A kind of accurate prognoses system learnt based on big data depth

[technical field]

The present invention relates to a kind of accurate prognoses system learnt based on big data depth, belong to technical field of electronic commerce.

[background technology]

The development of the Internet changes the life of people, and none is not closely bound up with it in our clothing, food, lodging and transportion--basic necessities of life, and matched is the flourish of all kinds of ecommerce. One effective e-commerce platform not only can set up a kind of merchandise display between businessman and end user, kinds of goods are bought and sold, the correlation functions such as user's payment, with greater need for accomplishing shopping guide's service of precisely recommending and rationalizing in user's purchase, thus improving user for the satisfaction of e-commerce platform and degree of adhesion.

One e-commerce platform that can realize precisely recommendation and rationalization guidance has following advantage: (1) Accurate Analysis is liked with grasping client, accomplish differentiated service and customized product promotion, and then help businessman to optimize commodity stocks and capital turnover; (2) by corresponding customer behavior analysis so that advertisement is more accurate, obtains best benefit ratio, by helping client to find emerging trend, it is to avoid client's dislike that invalid advertisement causes, customer loyalty and then the volume that boosts consumption are increased; (3) help client to find rapidly commodity interested, reduce invalid advertisement and related cost, promote unit and put into the economic benefit produced.

The method and system being analyzed according to electronic commerce data both at home and abroad at present is in the starting stage, but they utilize the database analysis system that data base itself has mostly. Such way is fairly simple in realization, but rate of precision is inadequate, particularly under the premise that big data exist at present. Along with the development of the reinforcement of hardware computing capability and degree of depth learning art, the depth analysis based on big data has had become as possibility, and realizes a set of prognoses system more accurately for this and also become more practical.

[summary of the invention]

It is an object of the invention to: for defect and the deficiency of prior art, provide a kind of accurate prognoses system learnt based on big data depth, this system is by collecting user interest and preference, thus providing best commercial product recommending and customizing recommendation, both help the commodity that user's fast searching is suitable, and also can be effectively improved the turnover of businessman.

For achieving the above object, the technical solution used in the present invention is:

A kind of accurate prognoses system learnt based on big data depth of the present invention, including following four parts:

(1), information extracting system: include data inputting module and database operating modules, according to the form arranged with businessman, the corresponding data of user basic information and transaction journal is provided by businessman, and the data that will provide for are analyzed and extract in the form that wherein important content writes data into data base by the operation of data base;

(2), training forecasting model system: include reconstructed module and training module, first it is that reconstructed module extracts data from above-mentioned database table, due in information extraction process, the information obtained is impaired and disappearance, according to restructing algorithm, the data of disappearance and redundancy in form are processed, rationally fill impaired and missing data accurately so that the data being trained can refine and representative more; After DSR, the training module of our design will be entered, it is achieved " on-line training pattern ", adapt dynamically to the magnanimity training data increased, improve training effectiveness, retrieve forecast model;

(3), active user inputs information system: include login module and input module, when certain user logs in, it is determined that its corresponding account and other essential information, input corresponding search condition with text mode, read its current querying condition, analyze its demand purpose;

(4), predictor system: include prediction module, according to the search condition provided according to user and the forecast model obtained in the training forecasting model system of (2nd) part, provide and recommend ranking accordingly, thus guiding user to do shopping more easily.

After adopting said method, present invention have the beneficial effect that the present invention will realize realizing the accurate analysis of big data based on the degree of depth learning art optimized, for client, it can find rapidly required commodity, reduces candidate item, finds new trend in time; For businessman, it can be promoted according to client interests, characterizes, and increases customer loyalty, increases sales volume, more understanding clients and market.

[accompanying drawing explanation]

Accompanying drawing described herein is used to provide a further understanding of the present invention, constitutes the part of the application, but is not intended that inappropriate limitation of the present invention, in the accompanying drawings:

Fig. 1 is the overall procedure schematic diagram of the present invention;

Fig. 2 is the training forecast model of the present invention;

Fig. 3 is the predictor system processing procedure in the present invention.

[detailed description of the invention]

Describe the present invention, illustrative examples therein and explanation in detail below in conjunction with accompanying drawing and specific embodiment to be only used for explaining the present invention, but not as a limitation of the invention.

As Figure 1-3, a kind of accurate prognoses system learnt based on big data depth, including following four parts:

(1), information extracting system: include data inputting module and database operating modules, according to the form arranged with businessman, the corresponding data of user basic information and transaction journal is provided by businessman, and the data that will provide for are analyzed and extract in the form that wherein important content writes data into data base by the operation of data base;

(2), training forecasting model system: include reconstructed module and training module, first it is that reconstructed module extracts data from above-mentioned database table, due in information extraction process, the information obtained is impaired and disappearance, according to restructing algorithm, the data of disappearance and redundancy in form are processed, rationally fill impaired and missing data accurately so that the data being trained can refine and representative more; After DSR, the training module of our design will be entered, it is achieved " on-line training pattern ", adapt dynamically to the magnanimity training data increased, improve training effectiveness, retrieve forecast model;

(3), active user inputs information system: include login module and input module, when certain user logs in, it is determined that its corresponding account and other essential information, input corresponding search condition with text mode, read its current querying condition, analyze its demand purpose;

(4), predictor system: include prediction module, according to the search condition provided according to user and the forecast model obtained in the training forecasting model system of (2nd) part, provide and recommend ranking accordingly, thus guiding user to do shopping more easily.

On-line prediction when electricity business's platform is in fact real by this system is as follows:

1, electricity business's platform provides data to be trained;

2, the backstage of electricity business's platform it is arranged on;

3 once there be user to input corresponding search condition, and electricity business's platform sends information to " accurate prognoses system "

" accurate prognoses system " provides recommendation list;

4, the electricity business website recommendation list according to feedback, prepares video data and is pushed to user.

The above is only the better embodiment of the present invention, therefore all equivalences done according to the structure described in present patent application scope, feature and principle change or modify, and are all included within the scope of present patent application.

Claims (1)

1. the accurate prognoses system learnt based on big data depth, it is characterised in that: include following four parts:

(1), information extracting system: include data inputting module and database operating modules, according to the form arranged with businessman, the corresponding data of user basic information and transaction journal is provided by businessman, and the data that will provide for are analyzed and extract in the form that wherein important content writes data into data base by the operation of data base;

(2), training forecasting model system: include reconstructed module and training module, first it is that reconstructed module extracts data from above-mentioned database table, due in information extraction process, the information obtained is impaired and disappearance, according to restructing algorithm, the data of disappearance and redundancy in form are processed, rationally fill impaired and missing data accurately so that the data being trained can refine and representative more; After DSR, the training module of our design will be entered, it is achieved " on-line training pattern ", adapt dynamically to the magnanimity training data increased, improve training effectiveness, retrieve forecast model;

(3), active user inputs information system: include login module and input module, when certain user logs in, it is determined that its corresponding account and other essential information, input corresponding search condition with text mode, read its current querying condition, analyze its demand purpose;

(4), predictor system: include prediction module, according to the search condition provided according to user and the forecast model obtained in the training forecasting model system of (2nd) part, provide and recommend ranking accordingly, thus guiding user to do shopping more easily.