CN105389718A - Automobile after-sale service recommendation method and system - Google Patents
Automobile after-sale service recommendation method and system Download PDFInfo
- Publication number
- CN105389718A CN105389718A CN201510896606.2A CN201510896606A CN105389718A CN 105389718 A CN105389718 A CN 105389718A CN 201510896606 A CN201510896606 A CN 201510896606A CN 105389718 A CN105389718 A CN 105389718A
- Authority
- CN
- China
- Prior art keywords
- automobile
- user
- service
- module
- mode sequences
- 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
Abstract
The present invention discloses an automobile after-sale service recommendation method and system. The system comprises a data acquisition module, an automobile basic data knowledge base building module, a pattern sequence generation module, a treelike decider building module, a user portrait building module, an after-sale service recommendation module and a physical store recommendation module, wherein the pattern sequence generation module automatically generates an extraction rule according to an automobile basic data knowledge base, and performs feature extraction and normalization on related discussed information of automobile after-sale services to enable each datum to be formalized into a pattern sequence; the user portrait building module adopts statistical information, hypothesis verification and significance analysis to obtain a plurality of distribution patterns of automobile after-sale service consumption of a user, so as to build a user portrait; the after-sale service recommendation module inputs a user vehicle feature set and the user portrait to a treelike decider, and recommends N paths with high matching degrees as service schemes to the user; and the physical store recommendation module recommends M optimal physical stores to the user for each automobile after-sale service.
Description
Technical field
The present invention relates to service recommendation field after automobile, particularly relate to a kind of based on service recommendation method and system after the automobile of large data.
Background technology
Society car owners get more and more, a lot of people selects self-driving to go to the airport, but, due to airport Limited parking and price is high, and periphery parking lot, airport is not because the reasons such as distance are utilized effectively, the car owner gone to the airport to self-driving brings larger puzzlement, stops and return the puzzlement of picking up the car in the airport that airport Dai Bo just in time solves car owner.Current airport generation pool service procedure is roughly: car owner before travel drives APP by micro-letter platform or generation and places an order, in system assignments generation, drives driver and parks for car owner in terminal wait according to car owner's designated time, during return, car owner is vehicle collection reservation again, and generation drives driver and opened by car to terminal and wait car owner to get.
At present, some Foreign Airport adopts the intelligent means such as robot to realize airport Dai Bo, and the generation pool service of domestic airport is also in Rapid development stage, but utilize large data and automobile basic data knowledge base for providing the technology of service recommendation after automobile immature after generation pool.The scope included is served very wide, service as more professional in self-service maintenance project (comprise and change machine oil and filter cleaner, air-conditioning system maintenance, brake system maintenance, windscreen wiper inspection and maintenance, car light inspection and maintenance, fuel system maintenance, engine interior maintenance, air inlet system and exhaust system maintenance etc.), car detailing and automobile major maintenance etc. after automobile.Therefore, we can to serve after the user on airport recommends suitable automobile to parking, and utilize and go on business during this period of time, allow the vehicle oneself berthed in airport be maintained to allow user, are to save time and laborsaving.But how could serve after recommending the automobile of, most worthy most suitable for its vehicle to different users suitable opportunity, be the still unsolved technical matters of industry so far.
Summary of the invention
Fundamental purpose of the present invention is based on large data mining technology, proposes service recommendation method and system after a kind of automobile, to serve after suitable opportunity recommends suitable automobile to user exactly.
The present invention is service recommendation method after the automobile that reaches above-mentioned purpose and provide, comprises the following steps:
Relevant discussion information is served after utilizing web crawlers technology to capture automobile;
According to the automobile basic data knowledge base built in advance, automatic generation decimation rule, described decimation rule is utilized to serve relevant discussion information after described automobile to carry out feature extraction, with service after producing many automobiles, data are discussed, and service discussion data after every bar automobile are standardized, and service after every bar automobile is discussed data all form turn to corresponding mode sequences, multiple mode sequences rock mechanism arrangement set;
Based on described mode sequences set, adopt the tree-shaped decision-making device of ID3 decision tree classifier technique construction;
Based on service after described many automobiles, data are discussed, statistical information, hypothesis verification and significance analysis is adopted to obtain the multiple distribution pattern of service consumption after user's automobile, all distribution patterns that probability of occurrence is greater than a threshold value are chosen from described multiple distribution pattern, form the user preferences vector of multiple correspondence, each user preferences vector and user basic information thereof form user's portrait;
Described user's portrait of the predesignated subscriber's vehicle characteristics set gathered in advance and correspondence is input in described tree-shaped decision-making device, matching degree height sequence according to each paths obtains TopN bar recommendation paths, after producing N kind automobile, service recommendation scheme is to the user with described predesignated subscriber vehicle characteristics set, simultaneously, serve for after every automobile in service recommendation scheme after N kind automobile, from service entities shop database after automobile, recommend TopM solid shop/brick and mortar store according to solid shop/brick and mortar store index weighting descending sort to user according to described mode sequences;
Wherein, M and N is natural number.
Service recommendation method after above-mentioned automobile provided by the invention, has the following advantages:
1) present invention utilizes the automobile basic data knowledge base of specialty, the automobile basic data knowledge base that this domain expert and practicing engineer provide determines high-quality and the accuracy of service recommendation after automobile to a great extent, also determines recommend method of the present invention and is different from conventional proposed algorithm.Conventional proposed algorithm comprises collaborative filtering, matrix decomposition, Rozman machine etc., what utilize is that user behavior data is to find parallel pattern, but everyone hobby is not unalterable, often grows with each passing hour, recommendation results and user preferences therefore can be caused not to be inconsistent.And the present invention not only sketches the contours with portrait according to user behavior, more crucially make use of described automobile basic data knowledge base, to grasp key information, make recommendation results more authoritative.If user behavior is very ripe, will find that recommendation of the present invention exceeds its expection completely, what have to a certain degree is intelligent.
2) present invention employs tree-shaped decision-making device, make full use of the information that large data contain, recommend TopN bar result to user.First tree-shaped decision-making device is with the obvious advantage in real world applications, is similar to human brain decision process most.Secondly, the preference of each user has difference, but large-scale crowd will manifest statistical law.Based on the tree-shaped decision-making device of large-scale data, the statistical law of large data can be made full use of, make the most reliable recommendation.
3) the structure let us of user's portrait is after the statistical property utilizing large data, fully can look after the tiny difference of each user preferences.The statistical law of large data likes different combinations with individual, provide not only high-quality recommendation, and give also personalized recommendation.
In preferred scheme, described automobile basic data knowledge base builds according to serving relevant basic expert data after automobile services encyclopaedia information and the artificial constructed automobile of expert.
In preferred scheme, before carrying out described feature extraction, first carry out filtration treatment to serving relevant discussion information after described automobile.
In preferred scheme, described feature extraction specifically comprises: extract keyword and value corresponding to keyword; Emotion word is extracted according to emotion word dictionary.
In preferred scheme, data are discussed to service after every bar automobile and standardizes and specifically comprise: emotion word is converted into value between-10 ~ 10 according to the polarity of emotion word dictionary definition and intensity; Keyword is converted into the corresponding title in described automobile basic data knowledge base, value corresponding for keyword is converted into measure value; Service after every bar automobile is discussed data all form turn to corresponding mode sequences and specifically adopt hidden markov model approach and Recognition with Recurrent Neural Network method.
In preferred scheme, described recommend method also comprises self study set-up procedure, and described self study set-up procedure specifically comprises:
The basis all joined by services consumption information after each user's automobile after building described user portrait and described automobile needed for the database of service entities shop inputs in data;
By services consumption information after each user's automobile all form turn to corresponding mode sequences and join in described mode sequences set, and utilize adaboost method to upgrade the weight of each feature on each mode sequences in described mode sequences set, to improve the Reliability of each mode sequences.By self study set-up procedure, make the change that recommend method provided by the invention can be accustomed to along with the drift of data characteristic and user preferences, the characteristic accurately after study to change, recommends result accurately in time.
In preferred scheme, described recommend method also comprise according to user browse footprint, search record, survey record and initiatively recommendation service carry out Collection and analysis user preferences, think the structure of described mode sequences set and described user portrait newly-increased basis input data.
For reaching above-mentioned purpose, the present invention separately provides service recommendation system after a kind of automobile, comprising: comprise data acquisition module, automobile basic data construction of knowledge base module, mode sequences generation module, tree-shaped decision-making device build module, user draws a portrait and builds module, rear service recommendation module and solid shop/brick and mortar store recommending module;
Described data acquisition module serves relevant discussion information after being used for capturing automobile by web crawlers technology;
Described mode sequences generation module is connected to described data acquisition module and described automobile basic data construction of knowledge base module, for the automobile basic data knowledge base constructed by described automobile basic data construction of knowledge base module, automatic generation decimation rule, serve relevant discussion information after described automobile to utilize described decimation rule and carry out feature extraction, thus data are discussed in service after producing many automobiles, and service discussion data after every bar automobile are standardized, and service after every bar automobile is discussed data all form turn to corresponding mode sequences, multiple mode sequences rock mechanism arrangement set,
Described tree-shaped decision-making device builds model calling to described mode sequences generation module, for according to described mode sequences set, adopts the tree-shaped decision-making device of ID3 decision tree classifier technique construction;
Described user draws a portrait and builds model calling to described mode sequences generation module, for discussing data according to service after described many automobiles, statistical information, hypothesis verification and significance analysis is adopted to obtain the multiple distribution pattern of service consumption after user's automobile, all distribution patterns that probability of occurrence is greater than a threshold value are chosen from described multiple distribution pattern, form the user preferences vector of multiple correspondence, each user preferences vector and user basic information thereof form user's portrait;
Described rear service recommendation model calling is drawn a portrait to described user and is built module and described tree-shaped decision-making device structure module, for being input in described tree-shaped decision-making device according to described user's portrait of the predesignated subscriber's vehicle characteristics set gathered in advance and correspondence, obtain TopN bar recommendation paths according to the matching degree of each paths, after producing N kind automobile, service recommendation scheme is to the user with described predesignated subscriber vehicle characteristics set;
Described solid shop/brick and mortar store recommending module is connected to described rear service recommendation module and described mode sequences generation module, for: serve for after every automobile in service recommendation scheme after N kind automobile, from service entities shop database after automobile, recommend TopM solid shop/brick and mortar store to user according to described mode sequences; Wherein, M and N is natural number.
In preferred scheme, described commending system also comprises self study adjusting module, be connected to described mode sequences generation module, for by services consumption information after each user's automobile all form turn to corresponding mode sequences and join in described mode sequences set, and utilize adaboost method to upgrade the weight of each feature on each mode sequences in described mode sequences set;
Also be connected to described user to draw a portrait and build module and described solid shop/brick and mortar store recommending module, for services consumption information after each user's automobile all being joined in the basis input data that build after described user portrait and described automobile needed for the database of service entities shop.
In preferred scheme, described commending system also comprises user preferences detecting module, be connected to described mode sequences generation module and described user and draw a portrait structure module, carry out Collection and analysis user preferences for browse footprint, search record, survey record and active recommendation service according to user, think the structure of described mode sequences set and described user portrait newly-increased basis input data.
Accompanying drawing explanation
Fig. 1 is the block diagram of service recommendation system after a kind of automobile of providing of the preferred embodiment of the present invention.
Embodiment
Below in conjunction with accompanying drawing with preferred embodiment the invention will be further described.
The specific embodiment of the present invention provides service recommendation system after a kind of automobile, for serving after suitable opportunity recommends suitable automobile to user, described commending system comprises data acquisition module, automobile basic data construction of knowledge base module, mode sequences generation module, tree-shaped decision-making device builds module, user draws a portrait structure module, rear service recommendation module and solid shop/brick and mortar store recommending module.
Described data acquisition module serves relevant discussion information after being used for capturing automobile by web crawlers technology.Such as, relevant discussion information is served after service relevant forum, blog, question answering system, social networks etc. capture described automobile after the automobile internet, NoSQL (database of non-relational) storage mode can be adopted to store, the large data mining under convenient employing distributed environment.Serve relevant discussion information after described automobile and comprise a lot of bar, such as, serve relevant discussion information comment on the automobile of net crawl from masses after and have 100,000, every bar at least comprises the solid shop/brick and mortar store, user's automobile essential information, consumption price, service evaluation etc. of rear COS, user basic information, consumption.
With reference to figure 1, described mode sequences generation module is connected to described data acquisition module and described automobile basic data construction of knowledge base module, for the automobile basic data knowledge base constructed by described automobile basic data construction of knowledge base module, automatic generation decimation rule, and utilize described decimation rule to serve relevant discussion information after described automobile to carry out feature extraction, thus data are discussed in service after producing many automobiles, and service discussion data after every bar automobile are standardized, and service after every bar automobile is discussed data all form turn to corresponding mode sequences, multiple mode sequences rock mechanism arrangement set.
Described automobile basic data construction of knowledge base module builds described automobile basic data knowledge base according to serving relevant basic expert data after automobile services encyclopaedia information and the artificial constructed automobile of expert.Described automobile basic data knowledge base comprises the relevant rear service content of concrete vehicle and price, and can cover many brands and the various thereof of domestic and imported, the accessory of covering, more than K ten thousand kinds, namely has K many ten thousand related datas at least.The parts price of every money vehicle, alternative accessory, working hour expense price, maintenance rule etc. are all put in storage.Cover the logical relation in professional domain between the node of automobile basic data knowledge base, comprise stratigraphic classification relation and example categorizing information.The fine granularity of automobile basic data knowledge base and rear service recommendation is closely related, and the result that fine granularity is recommended is exactly the ontology knowledge of knowledge base basecoat, serves after namely concrete automobile.
Preferably, before carrying out described feature extraction, first carrying out following filtration treatment to serving relevant discussion information after described automobile, after obtaining comparatively clean data, carrying out described feature extraction again.Described filtration treatment such as can comprise following process: first remove stop words, preliminary denoising sound; Recycling 2-10 rank language model removes noise further, and wherein the training data of language model is from described automobile basic data knowledge base; Then LDA sorter is adopted the non-data about automobile to be filtered out further again.
In a kind of specific embodiment, described feature extraction specifically comprises: extract keyword and value corresponding to keyword; Emotion word is extracted according to emotion word dictionary.Wherein, each node and the instance data of described automobile basic data knowledge base taken from keyword, such as, comprise service item, vehicle parameter, vehicle condition parameter, solid shop/brick and mortar store parameter etc. after automobile.
In a particular embodiment, to service after every bar automobile, data are discussed to standardize and specifically comprise: emotion word is converted into value between-10 ~ 10 according to the polarity (i.e. positive negativity) of emotion word dictionary definition and intensity; Keyword is converted into the corresponding title in described automobile basic data knowledge base, value corresponding for keyword is converted into measure value.Such as: credit scoring specification turns to the numeral between 1-10.
Service after every bar automobile is discussed data all form turn to corresponding mode sequences and specifically adopt hidden markov model approach and Recognition with Recurrent Neural Network method.Mode sequences can adopt (X, Y) → (S) → (Z) form represents, default item acquiescence item substitutes, wherein X representative of consumer vehicle characteristics set, Y representative of consumer hobby vector, S represents rear service plan, Z presentation-entity shop scheme, mode sequences (X, Y) → (S) → (Z) characterizes has the set of a certain user's vehicle characteristics and the rear service plan that the user meeting user preferences vector Y adopts is S, and solid shop/brick and mortar store is Z.The mode sequences that every bar data are formed forms mode sequences set jointly.
With reference to figure 1, described tree-shaped decision-making device builds model calling to described mode sequences generation module, for according to described mode sequences set, adopts the tree-shaped decision-making device of ID3 decision tree classifier technique construction.The step building tree-shaped decision-making device specifically comprises: the concept hierarchy first provided according to automobile basic data knowledge base with associate, construct the tree-shaped decision-making device model on basis, the hyponymy namely between the feature that uses of each node and node.Then utilize example information to train tree-shaped decision-making device model, described example information comes from the mode sequences set extracted according to automobile basic data knowledge base.The process of training finds the most suitable threshold value of feature on each node to classify exactly, and these threshold values will use the correlation computations of mutual information and gain entropy.Desired tree-shaped decision-making device just can be obtained through constantly training.
With reference to figure 1, described user draws a portrait and builds model calling to described mode sequences generation module, for discussing data according to service after described many automobiles, statistical information, hypothesis verification and significance analysis is adopted to obtain the multiple distribution pattern of service consumption after user's automobile, all distribution patterns that probability of occurrence is greater than a threshold value are chosen from described multiple distribution pattern, form the user preferences vector of multiple correspondence, each user preferences vector and user basic information thereof form user's portrait.At the system constructing initial stage, not yet there is service consumption after user's automobile of native system, serve relevant discussion information to obtain after the described automobile that then after user's automobile, the distribution pattern of service consumption then gathers based on data acquisition module, obtain user's portrait that a few class is general.Along with the operation of native system, service consumption after the automobile of system user can be joined user as feedback data and draw a portrait in the process of structure, the structure that user is drawn a portrait is more and more precisely with targeted.Described user preferences vector comprises sensu lato maintenance frequency be inclined to every rear service, and the quantification of user to the product corresponding to every rear service and solid shop/brick and mortar store is inclined to.
With reference to figure 1, described rear service recommendation model calling is drawn a portrait to described user and is built module and described tree-shaped decision-making device structure module, for being input in described tree-shaped decision-making device according to described user's portrait of the predesignated subscriber's vehicle characteristics set gathered in advance and correspondence, TopN is obtained (by the sequence of matching degree height according to the matching degree of each paths, select the front N bar that matching degree is higher) bar recommendation paths, after producing N kind automobile, service recommendation scheme is to the user with described predesignated subscriber vehicle characteristics set.The set of user's vehicle characteristics is airport generation pool system acquisition, such as after generation moors, carry out vehicle condition detection by practicing engineer, the vehicle condition report of concentrated expression vehicle condition is produced after detection, comprise maintenance prompting, average fuel consumption, driving score, fault and self-inspection information etc., storage enters system, as user's vehicle characteristics set of this user.Rear service recommendation process specifically comprises: be input in described tree-shaped decision-making device by user's portrait of user's vehicle characteristics set of a certain user and coupling, from the entrance of tree-shaped decision-making device, calculate the matching degree of every paths, find out the N paths that matching degree is the highest, namely this N paths represents service recommendation scheme after N kind automobile, recommend this user, wherein N is natural number.
With reference to figure 1, described solid shop/brick and mortar store recommending module is connected to described rear service recommendation module and described mode sequences generation module, for: serve for after every automobile in service recommendation scheme after N kind automobile, from service entities shop database after automobile, recommend TopM solid shop/brick and mortar store according to solid shop/brick and mortar store index weighting descending sort to user according to described mode sequences; M is natural number.The recommendation of solid shop/brick and mortar store is by technical quality, service quality, environmental quality three index weighting descending sorts.If user has the solid shop/brick and mortar store of hobby, then the solid shop/brick and mortar store of correspondence is added recommendation results, if in recommendation results, then strengthen recommended intensity.
Particularly, the step coming recommended entity shop according to mode sequences roughly comprises: based on the rear service plan of aforementioned recommendation, according to characteristic matching, find out matching degree the most much higher mode sequences, due to mode sequences, last includes the information (but do not get rid of have default item) of solid shop/brick and mortar store, according to feature directly and solid shop/brick and mortar store sets match obtain solid shop/brick and mortar store Candidate Set, again according to technical quality, service quality, environmental quality three index weighting descending sorts, thus obtain the highest M of a score solid shop/brick and mortar store and recommend user.
Recommended flowsheet is such as:
X → S: according to the maintenance rule in described automobile basic data knowledge base, system pretreatment portion branch makes a decision each accessory in user's vehicle characteristics set X, judges whether be greater than a predetermined time limit tenure of use; According to maintenance frequency and last time service time, judged whether new maintenance period; Whether function is normal, and whether outward appearance is normal, judges whether maintenance or more renews accessory, thus obtains the automobile presence information that quantizes, and then as the input of rear service recommendation module, through the automatic discrimination of tree-shaped decision-making device, obtains service plan.
Y → S: according to described user preferences vector Y, judges whether every service has arrived the new seeervice cycle, and basis for estimation is the service frequency F1 of user preferences, service time last time F2.Obtain service plan, merge with service plan obtained in the previous step, obtain final suggested design (comprising concrete rear service item and solid shop/brick and mortar store).
In a more preferred embodiment, with reference to figure 1, described commending system also comprises self study adjusting module, be connected to described mode sequences generation module, for by services consumption information after the user's automobile at every turn entering system all form turn to corresponding mode sequences and join in described mode sequences set, and utilize adaboost method to upgrade the weight of each feature on each mode sequences in described mode sequences set, to improve the Reliability of each mode sequences.Such as, in the operational process of native system, services consumption information after newly-increased user's automobile, then feature extraction, standardized operation are carried out to form a corresponding mode sequences to this information, and this mode sequences is joined in mode sequences set, upgrade the feature weight with this mode sequences with the mode sequences of same characteristic features simultaneously.
Described self study adjusting module is also connected to described user and draws a portrait and build module and described solid shop/brick and mortar store recommending module, for all being joined by services consumption information after each user's automobile in the basis input data that build after described user portrait and described automobile needed for the database of service entities shop.Such as, services consumption information after newly-increased user's automobile, then this information processing is become corresponding distribution pattern, form a user preferences vector, essential information again in conjunction with this user forms corresponding user's portrait, then sketching the contours of new user's portrait, then, when native system gives this user's recommendation service next time, will be more accurate.In a word, self study adjusting module sets up information feed back passage, constantly customer consumption information is added other modules and carries out data processing, then the rear service recommendation result finally obtained and recommendation opportunity will be more and more accurate.Separately, feedback information is also incorporated in the generative process of tree-shaped decision-making device in the mode of Bayes KL distance by self study adjusting module, Bayes KL distance is shown in formula 1, wherein pat refers to the concrete pattern (context model that pattern is here served after comprising automobile, such as car surface is dirty to needing carwash service during much degree, need to build tree-shaped decision-making device by the method for discrimination that mutual information is relevant with gain entropy), D finger ring environment information, feature on every paths all can be regarded as the one of environmental information, and the specific environment that every paths comprises can be different:
In another kind of preferred embodiment, with reference to figure 1, described commending system also comprises user preferences detecting module, be connected to described mode sequences generation module and described user and draw a portrait structure module, carry out Collection and analysis user preferences for browse footprint, search record, survey record and active recommendation service according to user, think the structure of described mode sequences set and described user portrait newly-increased basis input data.
Above content is in conjunction with concrete preferred implementation further description made for the present invention, can not assert that specific embodiment of the invention is confined to these explanations.For those skilled in the art, without departing from the inventive concept of the premise, some equivalent to substitute or obvious modification can also be made, and performance or purposes identical, all should be considered as belonging to protection scope of the present invention.
Claims (10)
1. a service recommendation method after automobile, is characterized in that, comprises the following steps:
Relevant discussion information is served after utilizing web crawlers technology to capture automobile;
According to the automobile basic data knowledge base built in advance, automatic generation decimation rule, described decimation rule is utilized to serve relevant discussion information after described automobile to carry out feature extraction, with service after producing many automobiles, data are discussed, and service discussion data after every bar automobile are standardized, and service after every bar automobile is discussed data all form turn to corresponding mode sequences, multiple mode sequences rock mechanism arrangement set; Wherein, described automobile basic data knowledge base
Based on described mode sequences set, adopt the tree-shaped decision-making device of ID3 decision tree classifier technique construction;
Based on service after described many automobiles, data are discussed, statistical information, hypothesis verification and significance analysis is adopted to obtain the multiple distribution pattern of service consumption after user's automobile, all distribution patterns that probability of occurrence is greater than a threshold value are chosen from described multiple distribution pattern, form the user preferences vector of multiple correspondence, each user preferences vector and user basic information thereof form user's portrait;
Described user's portrait of the predesignated subscriber's vehicle characteristics set gathered in advance and correspondence is input in described tree-shaped decision-making device, matching degree height sequence according to each paths obtains TopN bar recommendation paths, after producing N kind automobile, service recommendation scheme is to the user with described predesignated subscriber vehicle characteristics set, simultaneously, serve for after every automobile in service recommendation scheme after N kind automobile, from service entities shop database after automobile, recommend TopM solid shop/brick and mortar store according to solid shop/brick and mortar store index weighting descending sort to user according to described mode sequences;
Wherein, M and N is natural number.
2. service recommendation method after automobile as claimed in claim 1, is characterized in that: described automobile basic data knowledge base builds according to serving relevant basic expert data after automobile services encyclopaedia information and the artificial constructed automobile of expert.
3. service recommendation method after automobile as claimed in claim 1, is characterized in that: before carrying out described feature extraction, first carrying out filtration treatment to serving relevant discussion information after described automobile.
4. service recommendation method after automobile as claimed in claim 1, is characterized in that: described feature extraction specifically comprises: extract keyword and value corresponding to keyword; Emotion word is extracted according to emotion word dictionary.
5. service recommendation method after automobile as claimed in claim 4, is characterized in that: to service after every bar automobile, data are discussed and standardize and specifically comprise: emotion word is converted into value between-10 ~ 10 according to the polarity of emotion word dictionary definition and intensity; Keyword is converted into the corresponding title in described automobile basic data knowledge base, value corresponding for keyword is converted into measure value;
Service after every bar automobile is discussed data all form turn to corresponding mode sequences and specifically adopt hidden markov model approach and Recognition with Recurrent Neural Network method.
6. service recommendation method after automobile as claimed in claim 1, it is characterized in that: also comprise self study set-up procedure, described self study set-up procedure specifically comprises:
The basis all joined by services consumption information after each user's automobile after building described user portrait and described automobile needed for the database of service entities shop inputs in data;
By services consumption information after each user's automobile all form turn to corresponding mode sequences and join in described mode sequences set, and utilize adaboost method to upgrade the weight of each feature on each mode sequences in described mode sequences set, to improve the Reliability of each mode sequences.
7. service recommendation method after automobile as claimed in claim 1, it is characterized in that: also comprise and carry out Collection and analysis user preferences according to browse footprint, search record, survey record and the active recommendation service of user, think the structure of described mode sequences set and described user portrait newly-increased basis input data.
8. a service recommendation system after automobile, is characterized in that: comprise data acquisition module, automobile basic data construction of knowledge base module, mode sequences generation module, tree-shaped decision-making device builds module, user draws a portrait structure module, rear service recommendation module and solid shop/brick and mortar store recommending module;
Described data acquisition module serves relevant discussion information after being used for capturing automobile by web crawlers technology;
Described mode sequences generation module is connected to described data acquisition module and described automobile basic data construction of knowledge base module, for the automobile basic data knowledge base constructed by described automobile basic data construction of knowledge base module, automatic generation decimation rule, serve relevant discussion information after described automobile to utilize described decimation rule and carry out feature extraction, thus data are discussed in service after producing many automobiles, and service discussion data after every bar automobile are standardized, and service after every bar automobile is discussed data all form turn to corresponding mode sequences, multiple mode sequences rock mechanism arrangement set,
Described tree-shaped decision-making device builds model calling to described mode sequences generation module, for according to described mode sequences set, adopts the tree-shaped decision-making device of ID3 decision tree classifier technique construction;
Described user draws a portrait and builds model calling to described mode sequences generation module, for discussing data according to service after described many automobiles, statistical information, hypothesis verification and significance analysis is adopted to obtain the multiple distribution pattern of service consumption after user's automobile, all distribution patterns that probability of occurrence is greater than a threshold value are chosen from described multiple distribution pattern, form the user preferences vector of multiple correspondence, each user preferences vector and user basic information thereof form user's portrait;
Described rear service recommendation model calling is drawn a portrait to described user and is built module and described tree-shaped decision-making device structure module, for being input in described tree-shaped decision-making device according to described user's portrait of the predesignated subscriber's vehicle characteristics set gathered in advance and correspondence, matching degree height sequence according to each paths obtains TopN bar recommendation paths, and after producing N kind automobile, service recommendation scheme is to the user with described predesignated subscriber vehicle characteristics set;
Described solid shop/brick and mortar store recommending module is connected to described rear service recommendation module and described mode sequences generation module, for: serve for after every automobile in service recommendation scheme after N kind automobile, from service entities shop database after automobile, recommend TopM solid shop/brick and mortar store according to solid shop/brick and mortar store index weighting descending sort to user according to described mode sequences; Wherein, M and N is natural number.
9. service recommendation system after automobile as claimed in claim 8, it is characterized in that: also comprise self study adjusting module, be connected to described mode sequences generation module, for by services consumption information after each user's automobile all form turn to corresponding mode sequences and join in described mode sequences set, and utilize adaboost method to upgrade the weight of each feature on each mode sequences in described mode sequences set;
Also be connected to described user to draw a portrait and build module and described solid shop/brick and mortar store recommending module, for services consumption information after each user's automobile all being joined in the basis input data that build after described user portrait and described automobile needed for the database of service entities shop.
10. service recommendation system after automobile as claimed in claim 8, it is characterized in that: also comprise user preferences detecting module, be connected to described mode sequences generation module and described user and draw a portrait structure module, carry out Collection and analysis user preferences for browse footprint, search record, survey record and active recommendation service according to user, think the structure of described mode sequences set and described user portrait newly-increased basis input data.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510896606.2A CN105389718A (en) | 2015-12-07 | 2015-12-07 | Automobile after-sale service recommendation method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510896606.2A CN105389718A (en) | 2015-12-07 | 2015-12-07 | Automobile after-sale service recommendation method and system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN105389718A true CN105389718A (en) | 2016-03-09 |
Family
ID=55421976
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510896606.2A Pending CN105389718A (en) | 2015-12-07 | 2015-12-07 | Automobile after-sale service recommendation method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105389718A (en) |
Cited By (20)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106127316A (en) * | 2016-06-30 | 2016-11-16 | 大连楼兰科技股份有限公司 | Automated vehicle maintenance prediction system |
| CN106204791A (en) * | 2016-06-30 | 2016-12-07 | 大连楼兰科技股份有限公司 | Automated vehicle based on technology of Internet of things maintenance predictor method |
| CN106204793A (en) * | 2016-06-30 | 2016-12-07 | 大连楼兰科技股份有限公司 | Vehicle maintenance predictor method |
| CN106897404A (en) * | 2017-02-14 | 2017-06-27 | 中国船舶重工集团公司第七0九研究所 | A kind of recommendation method and system based on many GRU layers of neutral nets |
| CN108287871A (en) * | 2017-12-28 | 2018-07-17 | 广州道本信息科技有限公司 | A kind of auto repair data check service system |
| CN108460651A (en) * | 2018-01-04 | 2018-08-28 | 金瓜子科技发展（北京）有限公司 | Vehicle recommends method and device |
| CN108734297A (en) * | 2017-04-24 | 2018-11-02 | 微软技术授权有限责任公司 | The machine learning commending system of the performance optimization of electronic content items for network transmission |
| CN108920521A (en) * | 2018-06-04 | 2018-11-30 | 上海财经大学 | User's portrait-item recommendation system and method based on pseudo- ontology |
| CN108932644A (en) * | 2017-05-26 | 2018-12-04 | 车伯乐（北京）信息科技有限公司 | A kind of user selects vehicle method, apparatus, equipment and computer-readable medium |
| CN109300018A (en) * | 2018-10-31 | 2019-02-01 | 深圳市元征科技股份有限公司 | A kind of Vehicular intelligent recommended method, device, equipment and storage medium |
| CN110163723A (en) * | 2019-05-20 | 2019-08-23 | 深圳市和讯华谷信息技术有限公司 | Recommended method, device, computer equipment and storage medium based on product feature |
| CN110400155A (en) * | 2018-04-20 | 2019-11-01 | 比亚迪股份有限公司 | Vehicle service platform and the recommended method of automobile services, system |
| CN110889566A (en) * | 2018-08-21 | 2020-03-17 | 上海博泰悦臻网络技术服务有限公司 | Internet of vehicles server, vehicle and vehicle accessory maintenance service pushing method |
| CN111401586A (en) * | 2020-04-09 | 2020-07-10 | 东风小康汽车有限公司重庆分公司 | Method and device for generating vehicle maintenance recommendation information |
| CN111797305A (en) * | 2020-03-16 | 2020-10-20 | 北京京东尚科信息技术有限公司 | Vehicle condition data processing method and device |
| CN111984689A (en) * | 2020-08-21 | 2020-11-24 | 北京百度网讯科技有限公司 | Information retrieval method, device, equipment and storage medium |
| CN112107866A (en) * | 2020-09-28 | 2020-12-22 | 腾讯科技（深圳）有限公司 | User behavior data processing method, device, equipment and storage medium |
| CN112184388A (en) * | 2020-10-12 | 2021-01-05 | 上海燕汐软件信息科技有限公司 | Home decoration service ticket distribution method and device and storage medium |
| CN113204714A (en) * | 2021-03-23 | 2021-08-03 | 北京中交兴路信息科技有限公司 | User portrait based task recommendation method and device, storage medium and terminal |
| CN117422232A (en) * | 2023-10-13 | 2024-01-19 | 上海复通软件技术有限公司 | Vehicle owner and customer operation method and system based on AIGC |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101568901A (en) * | 2006-12-21 | 2009-10-28 | 3M创新有限公司 | Computer-assisted sales of after-market motor vehicle products |
| CN103729747A (en) * | 2014-01-24 | 2014-04-16 | 林帼瑜 | Engine personalized maintenance plan recommendation system and method and personalized maintenance system |
| CN103793787A (en) * | 2014-01-23 | 2014-05-14 | 南京邮电大学 | Car networking processing system and method |
| WO2014131185A1 (en) * | 2013-02-28 | 2014-09-04 | 观致汽车有限公司 | Method and apparatus for vehicle maintenance alert |
-
2015
- 2015-12-07 CN CN201510896606.2A patent/CN105389718A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101568901A (en) * | 2006-12-21 | 2009-10-28 | 3M创新有限公司 | Computer-assisted sales of after-market motor vehicle products |
| WO2014131185A1 (en) * | 2013-02-28 | 2014-09-04 | 观致汽车有限公司 | Method and apparatus for vehicle maintenance alert |
| CN103793787A (en) * | 2014-01-23 | 2014-05-14 | 南京邮电大学 | Car networking processing system and method |
| CN103729747A (en) * | 2014-01-24 | 2014-04-16 | 林帼瑜 | Engine personalized maintenance plan recommendation system and method and personalized maintenance system |
Non-Patent Citations (1)
| Title |
|---|
| 双茜: "基于智能网络平台的汽车售后主动服务研究", 《中国优秀硕士学位论文全文数据库经济与管理科学辑》 * |
Cited By (29)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106204791B (en) * | 2016-06-30 | 2019-04-05 | 大连楼兰科技股份有限公司 | Automated vehicle based on technology of Internet of things maintains predictor method |
| CN106204791A (en) * | 2016-06-30 | 2016-12-07 | 大连楼兰科技股份有限公司 | Automated vehicle based on technology of Internet of things maintenance predictor method |
| CN106204793A (en) * | 2016-06-30 | 2016-12-07 | 大连楼兰科技股份有限公司 | Vehicle maintenance predictor method |
| CN106127316A (en) * | 2016-06-30 | 2016-11-16 | 大连楼兰科技股份有限公司 | Automated vehicle maintenance prediction system |
| CN106204793B (en) * | 2016-06-30 | 2019-04-05 | 大连楼兰科技股份有限公司 | Vehicle maintenance predictor method |
| CN106897404A (en) * | 2017-02-14 | 2017-06-27 | 中国船舶重工集团公司第七0九研究所 | A kind of recommendation method and system based on many GRU layers of neutral nets |
| CN106897404B (en) * | 2017-02-14 | 2021-04-09 | 中国船舶重工集团公司第七0九研究所 | Recommendation method and system based on multi-GRU layer neural network |
| CN108734297A (en) * | 2017-04-24 | 2018-11-02 | 微软技术授权有限责任公司 | The machine learning commending system of the performance optimization of electronic content items for network transmission |
| CN108734297B (en) * | 2017-04-24 | 2023-07-28 | 微软技术许可有限责任公司 | Machine learning recommendation system, method for performance optimization of electronic content items |
| CN108932644A (en) * | 2017-05-26 | 2018-12-04 | 车伯乐（北京）信息科技有限公司 | A kind of user selects vehicle method, apparatus, equipment and computer-readable medium |
| CN108287871A (en) * | 2017-12-28 | 2018-07-17 | 广州道本信息科技有限公司 | A kind of auto repair data check service system |
| CN108460651A (en) * | 2018-01-04 | 2018-08-28 | 金瓜子科技发展（北京）有限公司 | Vehicle recommends method and device |
| CN110400155B (en) * | 2018-04-20 | 2022-12-09 | 比亚迪股份有限公司 | Vehicle service system and vehicle service recommendation method and system |
| CN110400155A (en) * | 2018-04-20 | 2019-11-01 | 比亚迪股份有限公司 | Vehicle service platform and the recommended method of automobile services, system |
| CN108920521B (en) * | 2018-06-04 | 2021-07-09 | 上海财经大学 | User portrait-project recommendation system and method based on pseudo ontology |
| CN108920521A (en) * | 2018-06-04 | 2018-11-30 | 上海财经大学 | User's portrait-item recommendation system and method based on pseudo- ontology |
| CN110889566A (en) * | 2018-08-21 | 2020-03-17 | 上海博泰悦臻网络技术服务有限公司 | Internet of vehicles server, vehicle and vehicle accessory maintenance service pushing method |
| CN109300018A (en) * | 2018-10-31 | 2019-02-01 | 深圳市元征科技股份有限公司 | A kind of Vehicular intelligent recommended method, device, equipment and storage medium |
| CN110163723A (en) * | 2019-05-20 | 2019-08-23 | 深圳市和讯华谷信息技术有限公司 | Recommended method, device, computer equipment and storage medium based on product feature |
| CN111797305A (en) * | 2020-03-16 | 2020-10-20 | 北京京东尚科信息技术有限公司 | Vehicle condition data processing method and device |
| CN111401586A (en) * | 2020-04-09 | 2020-07-10 | 东风小康汽车有限公司重庆分公司 | Method and device for generating vehicle maintenance recommendation information |
| CN111984689B (en) * | 2020-08-21 | 2023-07-25 | 北京百度网讯科技有限公司 | Information retrieval method, device, equipment and storage medium |
| CN111984689A (en) * | 2020-08-21 | 2020-11-24 | 北京百度网讯科技有限公司 | Information retrieval method, device, equipment and storage medium |
| CN112107866A (en) * | 2020-09-28 | 2020-12-22 | 腾讯科技（深圳）有限公司 | User behavior data processing method, device, equipment and storage medium |
| CN112184388A (en) * | 2020-10-12 | 2021-01-05 | 上海燕汐软件信息科技有限公司 | Home decoration service ticket distribution method and device and storage medium |
| CN112184388B (en) * | 2020-10-12 | 2024-02-13 | 上海燕汐软件信息科技有限公司 | Home decoration service list distribution method, device and storage medium |
| CN113204714A (en) * | 2021-03-23 | 2021-08-03 | 北京中交兴路信息科技有限公司 | User portrait based task recommendation method and device, storage medium and terminal |
| CN117422232A (en) * | 2023-10-13 | 2024-01-19 | 上海复通软件技术有限公司 | Vehicle owner and customer operation method and system based on AIGC |
| CN117422232B (en) * | 2023-10-13 | 2024-04-16 | 上海复通软件技术有限公司 | AIGC-based vehicle owner and customer operation method and system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105389718A (en) | Automobile after-sale service recommendation method and system | |
| CN110084151B (en) | Video abnormal behavior discrimination method based on non-local network deep learning | |
| CN105808732A (en) | Integration target attribute identification and precise retrieval method based on depth measurement learning | |
| CN108647691B (en) | Image classification method based on click feature prediction | |
| CN103812872B (en) | A kind of network navy behavioral value method and system based on mixing Di Li Cray process | |
| CN106570477A (en) | Vehicle model recognition model construction method based on depth learning and vehicle model recognition method based on depth learning | |
| CN106296312A (en) | Online education resource recommendation system based on social media | |
| CN107563426A (en) | A kind of learning method of locomotive operation temporal aspect | |
| CN108921602B (en) | User purchasing behavior prediction method based on integrated neural network | |
| CN109635010B (en) | User characteristic and characteristic factor extraction and query method and system | |
| CN106572001B (en) | A kind of dialogue method and system of intelligent customer service | |
| CN109359670B (en) | Individual association strength automatic detection method based on traffic big data | |
| CN103593474A (en) | Image retrieval ranking method based on deep learning | |
| CN109829733A (en) | A kind of false comment detection system and method based on Shopping Behaviors sequence data | |
| CN105045807A (en) | Data cleaning algorithm based on Internet trading information | |
| CN110532379A (en) | A kind of electronics information recommended method of the user comment sentiment analysis based on LSTM | |
| CN106897359A (en) | Internet information is collected and correlating method | |
| CN104702465A (en) | Parallel network flow classification method | |
| CN109658194A (en) | A kind of lead referral method and system based on video frequency tracking | |
| CN109858974A (en) | Automobile-used family identification model construction method and recognition methods are purchased | |
| CN109145301A (en) | Information classification approach and device, computer readable storage medium | |
| CN115099450A (en) | Family carbon emission monitoring and accounting platform based on fusion model | |
| CN110109902A (en) | A kind of electric business platform recommender system based on integrated learning approach | |
| CN115687760A (en) | User learning interest label prediction method based on graph neural network | |
| CN105890600A (en) | Subway passenger position inferring method based on mobile phone sensors |
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: 20160309 |
|
| RJ01 | Rejection of invention patent application after publication |