CN110019616A - A kind of POI trend of the times state acquiring method and its equipment, storage medium, server - Google Patents

A kind of POI trend of the times state acquiring method and its equipment, storage medium, server Download PDF

Info

Publication number
CN110019616A
CN110019616A CN201711260201.5A CN201711260201A CN110019616A CN 110019616 A CN110019616 A CN 110019616A CN 201711260201 A CN201711260201 A CN 201711260201A CN 110019616 A CN110019616 A CN 110019616A
Authority
CN
China
Prior art keywords
poi
trend
target poi
target
state
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN201711260201.5A
Other languages
Chinese (zh)
Other versions
CN110019616B (en
Inventor
杜逸康
龚剑
陈永全
泮华杰
成睿
张金晶
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Tencent Technology Shenzhen Co Ltd
Original Assignee
Tencent Technology Shenzhen Co Ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Tencent Technology Shenzhen Co Ltd filed Critical Tencent Technology Shenzhen Co Ltd
Priority to CN201711260201.5A priority Critical patent/CN110019616B/en
Publication of CN110019616A publication Critical patent/CN110019616A/en
Application granted granted Critical
Publication of CN110019616B publication Critical patent/CN110019616B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
    • G06F16/29—Geographical information databases
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/90—Details of database functions independent of the retrieved data types
    • G06F16/95—Retrieval from the web
    • G06F16/953—Querying, e.g. by the use of web search engines
    • G06F16/9535—Search customisation based on user profiles and personalisation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/90—Details of database functions independent of the retrieved data types
    • G06F16/95—Retrieval from the web
    • G06F16/953—Querying, e.g. by the use of web search engines
    • G06F16/9537—Spatial or temporal dependent retrieval, e.g. spatiotemporal queries

Abstract

The embodiment of the invention discloses a kind of POI trend of the times state acquiring methods, which comprises obtains and the associated network information characteristic of target POI；The network information characteristic that will acquire is converted into corresponding trend of the times feature vector；The trend of the times feature vector for the target POI that conversion is obtained inputs trend of the times state judgment models, obtains the trend of the times state of target POI；Export the trend of the times state of the target POI.Using the present invention, the POI Up-to-date state in electronic map data can effectively ensure that.

Description

A kind of POI trend of the times state acquiring method and its equipment, storage medium, server
Technical field
The present invention relates to Internet technical field more particularly to a kind of POI trend of the times state acquiring method and its equipment, storage Medium, server.
Background technique
Electronic map, i.e. numerical map are the map for storing and consulting in a digital manner using computer technology, electronics Corresponding POI (Point of Interest, i.e. point of interest or information point) is shown to user according to different location in map.Ground The Up-to-date state of figure refers to that geospatial information provided by map will reflect current newest situation as much as possible, fast in city Under the new situation, the trend of the times state change of POI is very frequent, therefore the Up-to-date state of POI and accuracy are to electronic map for speed development It services most important.
The method being updated to POI trend of the times state mainly includes manually acquiring/reporting, and needs a large amount of manpower and material resources, It updates efficiency and timeliness is very undesirable.
Summary of the invention
In view of this, the embodiment of the present invention provides a kind of POI trend of the times state acquiring method, electronic map number can effectively ensure that POI Up-to-date state in.
In order to solve the above-mentioned technical problem, the embodiment of the invention provides a kind of POI trend of the times state acquiring method, the sides Method includes:
It obtains and the associated network information characteristic of target POI；
The network information characteristic that will acquire is converted into corresponding trend of the times feature vector；
The trend of the times feature vector for the target POI that conversion is obtained inputs trend of the times state judgment models, obtains showing for target POI Gesture state；
Export the trend of the times state of the target POI.
Correspondingly, the embodiment of the invention also provides a kind of POI trend of the times state obtaining apparatus, the equipment includes:
The network information obtains module, for obtaining and the associated network information characteristic of target POI；
Trend of the times feature conversion module, the network information characteristic for will acquire be converted into corresponding trend of the times feature to Amount；
The trend of the times feature vector input trend of the times state of trend of the times condition judgment module, the target POI for obtaining conversion is sentenced Disconnected model, obtains the trend of the times state of target POI；
Trend of the times state output module, for exporting the trend of the times state of the target POI.
Correspondingly, the embodiment of the invention also provides a kind of computer storage medium, the computer storage medium storages There is a plurality of instruction, described instruction is suitable for being loaded by processor and executing following steps:
It obtains and the associated network information characteristic of target POI；
The network information characteristic that will acquire is converted into corresponding trend of the times feature vector；
The trend of the times feature vector for the target POI that conversion is obtained inputs trend of the times state judgment models, obtains showing for target POI Gesture state；
Export the trend of the times state of the target POI.
Correspondingly, the embodiment of the invention also provides a kind of servers, comprising: processor and memory；Wherein, described to deposit Reservoir is stored with computer program, and the computer program is suitable for being loaded by the processor and executing following steps:
It obtains and the associated network information characteristic of target POI；
The network information characteristic that will acquire is converted into corresponding trend of the times feature vector；
The trend of the times feature vector for the target POI that conversion is obtained inputs trend of the times state judgment models, obtains showing for target POI Gesture state；
Export the trend of the times state of the target POI.
The embodiment of the present invention is realized fast according to the associated network information characteristic of target POI and trend of the times state judgment models Speed judges the trend of the times state of target POI, it is no longer necessary to inefficient hand data collection, the process for reporting and auditing, to have Effect ensure that the POI Up-to-date state in electronic map data.
Detailed description of the invention
In order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, to embodiment or will show below There is attached drawing needed in technical description to be briefly described, it should be apparent that, the accompanying drawings in the following description is only this Some embodiments of invention for those of ordinary skill in the art without creative efforts, can be with It obtains other drawings based on these drawings.
Fig. 1 is the implement scene configuration diagram of the POI trend of the times state acquiring method of the embodiment of the present invention；
Fig. 2 is a kind of flow diagram of POI trend of the times state acquiring method provided in an embodiment of the present invention；
Fig. 3 is that a kind of trend of the times state judgment models and at least trend of the times state judge between submodel in the embodiment of the present invention Relation schematic diagram；
Fig. 4 be the embodiment of the present invention on the electronic map displaying target POI when and meanwhile show the trend of the times shape of the target POI The schematic diagram of state；
Fig. 5 is that the embodiment of the present invention submits the query result of POI retrieval request to be changed because of POI trend of the times state in search box The schematic diagram of change；
Fig. 6 is the POI trend of the times state acquiring method flow diagram of another embodiment of the present invention；
Fig. 7 is the implementation logical framework schematic diagram of another embodiment of the present invention；
Fig. 8 is the POI trend of the times state acquiring method flow diagram of further embodiment of this invention；
Fig. 9 is the structural schematic diagram of the POI trend of the times state obtaining apparatus in the embodiment of the present invention；
Figure 10 is a kind of structural schematic diagram of server provided in an embodiment of the present invention.
Specific embodiment
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete Site preparation description, it is clear that described embodiments are only a part of the embodiments of the present invention, instead of all the embodiments.It is based on Embodiment in the present invention, it is obtained by those of ordinary skill in the art without making creative efforts every other Embodiment shall fall within the protection scope of the present invention.
POI trend of the times state acquiring method provided in an embodiment of the present invention can be applied to the POI trend of the times in electronic map data The mark of state or update, so that the electronic map data that user when obtaining geographic information services by electronic map, obtains It is more accurate credible.
The present embodiments relate to POI trend of the times state acquiring method execution depend on computer program, POI can be based on Trend of the times state obtaining apparatus is run on the computer system of Feng Ruoyiman system.The POI trend of the times state obtaining apparatus can be with It can also include tablet computer, personal computer (PC), intelligence including having the server of the functions such as storage, calculating and test The server apparatus such as energy mobile phone, palm PC and mobile internet device (MID, Mobile Internet Device).
Below in conjunction with attached drawing 1- attached drawing 8, it is situated between in detail to road condition information acquisition method provided in an embodiment of the present invention It continues.
It referring to Figure 1, is the implement scene configuration diagram of the POI trend of the times state acquiring method of the embodiment of the present invention, POI Trend of the times state obtaining apparatus can obtain the associated network information characteristic of POI from multiple network information characteristic providers According to, then handle to obtain the trend of the times state of POI according to the associated network information characteristic of POI, it on the one hand can be by electricity The trend of the times state of sub- map data base output POI, provides a user electronic map data, electronic map by electronic map database The user behavior that user is generated during using electronic map can also be supplied to the acquisition of POI trend of the times state and set by database It is standby；Another aspect POI trend of the times state obtaining apparatus can have electronic map data with self maintained, and itself provides a user electricity Sub- map datum simultaneously obtains the user behavior that user is generated during using electronic map.
Fig. 2 is referred to, for the embodiment of the invention provides a kind of flow diagrams of POI trend of the times state acquiring method.Such as Shown in Fig. 2, the embodiment of the present invention the method may include following steps S101- step S104.
S101 is obtained and the associated network information characteristic of target POI.
In embodiments of the present invention, target POI can be any POI or part POI or all in electronic map data POI, and the associated network information characteristic of target POI may include the access number of wireless access point corresponding with target POI According to, with the associated networks congestion control data of target POI or with the associated network media data of target POI.
The access data of the corresponding wireless access point of target POI can be the nothing for establishing incidence relation with target POI in advance The accessing user's quantity of line access point whithin a period of time, such as accessing user recently in several days is total or nearest several days Interior daily accessing user's quantity, several days can be 2-15 days.In the specific implementation, POI trend of the times state obtaining apparatus can To collect the access data of multiple wireless access point in a period of time, then in conjunction with multiple wireless access point for getting in advance with Incidence relation between POI, it is corresponding to obtain the access data of the corresponding wireless access point of each POI, wireless access point and POI it Between incidence relation can be reported by POI corresponding merchant, can also by manually acquire or third party's data providing provide, some POI can correspond to multiple wireless access point, such as one or more employee's access points, one or more customer's access points or backstage Office access point etc..The access data of the corresponding wireless access point of target POI can reflect the stream of people of target POI to a certain extent Situation, so as to be used to judge the trend of the times state of target POI.
It may include being directed to institute by recording user whithin a period of time with the associated networks congestion control data of target POI The data that the multiple network user behavior of target POI implementation obtains are stated, the multiple network user behavior includes click behavior, leads Navigation is, payment behavior, reservation behavior, comment behavior, splitting glass opaque and the behavior that reports an error etc., such as user is on the electronic map Be directed to the click of target POI input after retrieving POI or choose operation, check operation, navigation operation etc., it is this kind of to be directed to target The networks congestion control that POI is implemented can reflect to a certain extent for the degree of concern to the POI, so as to be used to judge The trend of the times state of target POI.The user behavior that described a period of time records can obtain to count in nearest several days User behavior, several days can be with such as 2-15 days.
Network media data can be all kinds of media datas occurred in network, such as information network's number of pages in web page media According to, data such as the information data of news category APP, the article of social network media, dynamic, message, POI trend of the times state obtaining apparatus A large amount of network media data can be got from the above media channel, and then be screened out from it and the associated network of POI Media data, and therefrom further analysis obtains media fragment relevant to POI trend of the times state, for judging the trend of the times shape of POI State, such as news item content are certain one floor Pizza Hut of mansion finishing, then the relevant target POI of this news data is exactly ground Location is located at the Pizza Hut of one floor of mansion, this news data obviously can be used as the associated network media data of target POI.
In an alternative embodiment, POI trend of the times state obtaining apparatus can also according between target POI and other POI according to Attached relationship, using there are the network information characteristics of other POI of the relations of dependence as the associated net of target POI with target POI Network information characteristics data may further include and the target POI phase with the associated network information characteristic of target POI The network information characteristic of pass and there are the relevant network information characteristics of other POI of the relations of dependence to the target POI According to.And the relevant network information characteristic of target POI is the access data and mesh of wireless access point corresponding with target POI Mark the associated networks congestion control data of POI or with the associated network media data of target POI, will exist with the target POI Other POI of the relations of dependence are described as the first POI, and there are the relevant networks of other POI of the relations of dependence to believe to the target POI Cease the access data and the associated networks congestion control number of the first POI that characteristic is wireless access point corresponding with the first POI According to or with the first associated network media data of POI.Such as the dining room A is located in the mansion B or the dining room C is located in the park D, that It there is the relationship mutually depended between the two POI, it, can be by the net of the latter when judging the trend of the times state of the former POI Network information characteristics data are taken into account together as reference.It exists simultaneously and is only deposited in the POI of unique dependence, such as the mansion E In a F large size house ornamentation supermarket, then network information characteristic any in the mansion E and the two POI of F supermarket can be made For judge another POI trend of the times state reference.
The relations of dependence between the POI manually can acquire or collect POI address information and analyze to obtain, can also basis Media text fragments simultaneously comprising the target POI title and other POI titles carry out natural language processing, and analysis obtains institute The relations of dependence between target POI and other POI are stated, such as is repeatedly occurring " dining room A in the mansion B " in news or " is going The text fragments of the similar contents such as past 2 floor of the mansion the B arrival dining room A ", can determine the two POI after semantic analysis is handled Between there is the relationship that mutually depends on.
In an alternative embodiment, POI trend of the times state obtaining apparatus can be special according to the associated network information of the POI got Sign data establish POI trend of the times information characteristics library, to read target POI association from POI trend of the times information characteristics library when needed Network information characteristic.
S102, the network information characteristic that will acquire are converted into corresponding trend of the times feature vector.
The process that network information characteristic is converted into corresponding trend of the times feature vector is considered as at a kind of normalization The process of reason.According to above-mentioned each network information characteristic got or at least two network information characteristics In conjunction with the trend of the times feature vector value that can determine target POI different dimensions, the trend of the times feature vector value group of all dimensions of setting At the trend of the times feature vector of target POI, if POI trend of the times state obtaining apparatus gets certain network information characteristic, then The trend of the times feature vector value of corresponding dimension can be determined according to the network information characteristic that gets, in the specific implementation, can be with Determine that each network is believed by the feature vector value comparison table of preset network information characteristic and respective dimensions Cease the corresponding feature vector value of characteristic, for following table 1, the corresponding feature of available 8 kinds of network information characteristics Vector value, (- 1,1, -1,0,0,1,1,0) can be the trend of the times feature vector of a target POI actually obtained, at other In alternative embodiment, the trend of the times feature vector of target POI can be determined using part contrast relationship therein, or in the table of comparisons Add the trend of the times feature vector that more contrast relationships are used to determine target POI.If POI trend of the times state obtaining apparatus simultaneously has not been obtained To certain network information characteristic of target POI, then this kind of network information characteristic can be corresponded to the trend of the times of dimension Feature vector value is set as a default value, such as 0.
Table 1: network information characteristic and feature vector value comparison table
In an alternative embodiment, if get with the associated network information characteristic of target POI further comprise with The relevant network information characteristic of the target POI and there are the relevant nets of other POI of the relations of dependence to the target POI Network information characteristics data, then similarly can be according to there are the relevant networks of other POI of the relations of dependence to believe to the target POI Breath characteristic obtains that corresponding there are the trend of the times feature vectors of other POI of the relations of dependence with the target POI.
The trend of the times feature vector of S103, the target POI that conversion is obtained input trend of the times state judgment models, obtain target The trend of the times state of POI.
The trend of the times feature vector input trend of the times state judgement for the target POI that POI trend of the times state obtaining apparatus obtains conversion Model, so that it may which judgement obtains the trend of the times state of target POI.
It should be noted that in the embodiment of the present application, trend of the times state judgment models are to have marked trend of the times shape according to multiple What the POI of state and the trend of the times feature vector corresponding with each POI for having marked trend of the times state got were trained, Multiple corresponding trend of the times states of POI for having marked trend of the times state are training sample.And it is described with each marked trend of the times state The corresponding trend of the times feature vector of POI, can use mode identical with the present embodiment S101-S102, marked according to each The network information characteristic of the POI of trend of the times state converts to obtain.It is understood that sentencing by constantly trained trend of the times state The trend of the times feature vector of multiple sample POI is substituted into the judging result that trend of the times state judgment models carry out trend of the times state by disconnected model Judging result, is finally most approached showing for sample POI annotation results by the annotation results that can constantly approach the multiple sample POI Gesture state judgment models are determined as the trend of the times state judgment models that training obtains.
In the specific implementation, trend of the times state judgment models can use GBDT (Gradient Boosting Decision Tree, gradient promote decision tree), XGboost machine learning function library or DNN (Deep Neural Networks, depth nerve Network) etc. machine learning models.
In an alternative embodiment, the trend of the times state judgment models can judge that submodel melts by least two trend of the times states Conjunction obtains, wherein each trend of the times state judges that submodel is corresponding at least one trend of the times feature vector.At least two display State judges that submodel can be combined with a fusion function to constitute trend of the times state judgment models, wherein the fusion function can To include multiple fusion parameters, each trend of the times state judges that submodel can correspond to a fusion in trend of the times state judgment models Parameter.The trend of the times state judges that submodel equally can be using appointing in the machine learning models such as GBDT, XGboost or DNN It is one or more.Trend of the times state judgment models and at least two trend of the times states judge that the relationship between submodel can be such as Fig. 3 institute Show, trend of the times state judgment models F (x) judges submodel F by multiple trend of the times statesn(xn) and it is sub with the judgement of each trend of the times state The corresponding fusion parameters a of modelnThe fusion function of composition collectively constitutes, and the mode of composition can be linear, or non- Linear, it is illustrative as follows:
F (x)=a1·F1(x1)+a2·F2(x2)+a3·F3(x3).....+an·Fn(xn)
Wherein F (x) represents trend of the times state judgment models, Fn(xn) indicate the trend of the times feature vector and correspondence of some dimension Trend of the times state judge submodel, anIt indicates to judge the corresponding fusion parameters of submodel with the trend of the times state.
POI trend of the times state obtaining apparatus can be according to multiple POI (i.e. sample POI) for having marked trend of the times state and acquisition To each POI for having marked trend of the times state trend of the times feature vector, submodel, which is trained, to be judged to corresponding trend of the times state, It is understood that judging submodel by constantly trained trend of the times state, the trend of the times feature vector of multiple sample POI is substituted into Trend of the times state judges that the judging result of submodel progress trend of the times state can constantly approach the mark knot of the multiple sample POI Fruit, the trend of the times state that judging result is finally most approached to sample POI annotation results judge that submodel is determined as training what is obtained to show Gesture state judges submodel.
Then POI trend of the times state obtaining apparatus is according to multiple POI for having marked trend of the times state, having marked with each of getting The corresponding trend of the times feature vector of POI and trained at least two trends of the times state for infusing trend of the times state judge submodule Type judges that the fusion parameters of submodel are trained to each trend of the times state is corresponded in the trend of the times state judgment models.To Judge that submodel judges submodule with trained corresponding each trend of the times state according to trained at least two trends of the times state The fusion parameters of type form trained trend of the times state judgment models.
S104 exports the trend of the times state of the target POI.
Specifically, POI trend of the times state obtaining apparatus can mark the trend of the times shape of the target POI in electronic map data State, so that electronic map when showing the target POI, can show the trend of the times state of the target POI to user simultaneously Make prompt.Such as shown in Fig. 4, user is chosen specific when opening electronic map by click or POI search frame on map After POI journal duck blood fans soup (Yongding lane shop), electronic map shows the relevant information of the POI, at this moment can be by the POI's Trend of the times state is prompted (" trade company's closing " item of information in figure) to user, additionally can for example be moved to cursor in user When above target POI, i.e., by the relative information displaying comprising POI trend of the times state to user.
In another embodiment, POI trend of the times state obtaining apparatus is when carrying out POI inquiry according to POI retrieval request, if looking into Asking in the multiple POI for meeting the POI retrieval request includes the target POI, then carries in the POI query result of return Other POI of the target POI are excluded in the multiple POI, or the target POI exists in the POI query result of return Display sequence in the multiple POI is arranged last.Such as shown in Fig. 5, user inputs " journal duck blood fans in search box The search condition of soup " submits POI retrieval request to POI trend of the times state obtaining apparatus, and POI trend of the times state obtaining apparatus is according to this Retrieval request carries out POI inquiry in electronic map data, and target POI journal duck blood fans soup (Yongding lane shop) also meets the inspection Therefore rope condition is also queried to, but target POI due to trend of the times situation be " closing ", can be as shown in figure 5, searching for Target POI is excluded in the POI query result that the lower section of frame provides, other POI for meeting the search condition are only provided, at other The display sequence of target POI can also be arranged in all POI's finally, can also be for meeting search condition in embodiment The trend of the times state of target POI is directly displayed in POI query result.The feedback result for changing POI inquiry through the above way, can be with It effectively avoids user from finding trend of the times state on map as closing or not available POI and therefore brings inconvenience.
In other alternative embodiments, POI trend of the times state obtaining apparatus can also be by electronic map database or electricity Sub- map datum provider exports the trend of the times state of the target POI, and the latter is allowed to be responsible in electronic map data described in mark The trend of the times state of target POI or the feedback result for changing POI inquiry.
POI trend of the times state obtaining apparatus in the present embodiment according to the associated network information characteristic of target POI and shows Gesture state judgment models can be realized and quickly judge the trend of the times state of target POI, it is no longer necessary to inefficient hand data collection, The process for reporting and auditing, thus the POI Up-to-date state being effectively ensured in electronic map data.
Fig. 6 and Fig. 7 are referred to, Fig. 7 is a kind of POI trend of the times state acquiring method provided in another embodiment of the present invention Implement logical framework schematic diagram, Fig. 6 is the POI trend of the times state acquiring method flow diagram implemented under the logical framework.Such as Shown in Fig. 6, the embodiment of the present invention the method may include following steps S201- step S208.
S201 obtains the trend of the times state of multiple sample POI marks.
Multiple sample POI, that is, multiple POI for having marked trend of the times state, mark trend of the times state can by manually marking, It can also be provided by third party.
S202 obtains the associated network information characteristic of multiple sample POI.
As shown in fig. 7, with the associated network information characteristic of sample POI including sample POI corresponding in the present embodiment WIFI accesses data, user's click behavioral data and user comment data for sample POI.In other alternative embodiments In, it may include more various contents with the associated network information characteristic of sample POI, can specifically refer to and implement above The description as described in network information characteristic in S101 in example repeats no more in the present embodiment.
The network information characteristic of sample POI is converted corresponding trend of the times feature vector by S203.
According to the knot of above-mentioned each network information characteristic got or at least two network information characteristics Close the trend of the times feature vector value that can determine sample POI different dimensions, the trend of the times feature vector value composition of all dimensions of setting The trend of the times feature vector of sample POI, if POI trend of the times state obtaining apparatus gets certain network information characteristic, then can To determine the trend of the times feature vector value of corresponding dimension according to the network information characteristic got, in the specific implementation, can lead to The feature vector value comparison table of preset network information characteristic and respective dimensions is crossed to determine each network information The corresponding feature vector value of characteristic.
S204, according to the trend of the times state of multiple sample POI mark and conversion obtain the trend of the times feature of multiple sample POI to Amount, is trained trend of the times state judgment models.
In the specific implementation, trend of the times state judgment models can be using machine learning models such as GBDT, XGboost or DNN.It is right Trend of the times state judgment models, which are trained, can actually regard a model parameter progress to trend of the times state judgment models as not The process of disconnected optimization, it is to be understood that by constantly trained trend of the times state judgment models, conversion is obtained into multiple samples The judging result that the trend of the times feature vector of POI substitutes into trend of the times state judgment models progress trend of the times state can constantly approach multiple samples The trend of the times state outcome of this POI mark, the trend of the times state that judging result is finally most approached to training sample annotation results judge mould Type is determined as the trend of the times state judgment models that training obtains.
In an alternative embodiment, if trend of the times state judgment models as shown in Figure 3 by multiple trend of the times states judge submodel with And the fusion function for each trend of the times state judging that the corresponding fusion parameters of submodel are constituted collectively constitutes, then POI trend of the times shape State obtains equipment can be according to multiple sample POI trend of the times state marked and the trend of the times feature of each sample POI got Vector judges that submodel is trained to corresponding trend of the times state, it is to be understood that sentences by constantly trained trend of the times state The trend of the times feature vector of multiple sample POI is substituted into trend of the times state and judges that submodel carries out the judgement of trend of the times state by disconnected submodel Judging result, is finally most approached sample POI annotation results by the annotation results that as a result can constantly approach the multiple sample POI Trend of the times state judge that submodel is determined as the obtained trend of the times state of training and judges submodel.Then POI trend of the times state acquisition is set For according to multiple POI for having marked trend of the times state, the trend of the times feature corresponding with each POI for having marked trend of the times state got Vector and trained at least two trends of the times state judge submodel, to corresponding in the trend of the times state judgment models Each trend of the times state judges that the fusion parameters of submodel are trained.To be sentenced according to trained at least two trends of the times state Disconnected submodel judges that the fusion parameters of submodel form trained trend of the times shape with trained corresponding each trend of the times state State judgment models.
The above S201-S204 is the training process of trend of the times state judgment models, which can be with subsequent S205-S208 The mutually indepedent completion of description, can also successively execute according to the sequence of the present embodiment.Wherein training process can be completed offline, It can also be according to the real-time annotation results for determining or updating obtained sample POI and the network information characteristic of sample POI It carries out on-line training and continuous retraining and optimization is carried out to trend of the times state judgment models in real time.
S205 obtains the associated network information characteristic of target POI.
Target POI in the present embodiment is all POI in specific POI set, i.e., executes this implementation according to predetermined period S205-S207 in example can once update the trend of the times state of all POI in specific POI set, to send out in time There is the POI changed in existing trend of the times state.
In the present embodiment as shown in Figure 6 includes target POI corresponding with the associated network information characteristic of target POI WIFI accesses data, user's click behavioral data and user comment data for target POI.In other alternative embodiments In, it may include more various contents with the associated network information characteristic of sample POI, can specifically refer to and implement above The description as described in network information characteristic in S101 in example repeats no more in the present embodiment.
The network information characteristic of target POI is converted corresponding trend of the times feature vector by S206.
According to the knot of above-mentioned each network information characteristic got or at least two network information characteristics Close the trend of the times feature vector value that can determine target POI different dimensions, the trend of the times feature vector value composition of all dimensions of setting The trend of the times feature vector of target POI, if POI trend of the times state obtaining apparatus gets certain network information characteristic, then can To determine the trend of the times feature vector value of corresponding dimension according to the network information characteristic got, in the specific implementation, can lead to The feature vector value comparison table of preset network information characteristic and respective dimensions is crossed to determine each network information The corresponding feature vector value of characteristic.
The trend of the times feature vector of target POI is inputted trained trend of the times state judgment models, obtains target by S207 The trend of the times state of POI.
S208, by the trend of the times state output of target POI to application scenarios.
The application scenarios can such as Fig. 4 or electronic map application scenarios shown in fig. 5, can also be for example by trend of the times shape The changed POI of state is disclosed or is broadcasted, or is changed user's stroke planning according to the POI that trend of the times state change occurs or led Air route line etc..
The present embodiment by using multiple sample POI trend of the times state marked and the trend of the times feature of multiple sample POI to Amount trend of the times state judgment models are trained, so as to by trained trend of the times state judgment models to target POI Trend of the times state judged, it is no longer necessary to inefficient hand data collection, the process for reporting and auditing, to be effectively ensured POI Up-to-date state in electronic map data.
Fig. 8 is referred to, is a kind of POI trend of the times state acquiring method process signal provided in another embodiment of the present invention Figure.As shown in figure 8, the embodiment of the present invention the method may include following steps S301- step S309.
S301 collects the network media text information in first time period.
First time period in the present embodiment can be nearest 2 weeks, 3 weeks or nearest one month time window.POI trend of the times shape State obtains equipment can be by collecting the information web data in web page media, the information data of news category APP, social networks matchmaker The medias channel such as article, dynamic, message of body gets a large amount of network media text information.In an alternative embodiment, POI trend of the times state obtaining apparatus can be collected into network media text information according to the setting period again.
S302, according to full dose POI name set and first instance identification model from the network media text being collected into Media text fragments relevant to POI title are extracted in information.
Full dose POI name set in the present embodiment can be all POI set in electronic map data, i.e. POI is existing Gesture state obtaining apparatus is according to all POI titles in electronic map data, from the network matchmaker in the first time period being collected into It is filtered out in body text information and the associated media text fragments of POI.
The first instance identification model include AC automatic machine, condition random field entity recognition model (CRF, Conditional random field) or two-way shot and long term memory condition random field neural network model (BI-LSTM-CRF, Bi-directional-Long Short-Term Memory-CRF), that is, any one of mode is included the following three types from collection To first time period in network media text information in filter out and the associated media text fragments of POI:
1) by AC automaton model, the network media that all POI name-matches in electronic map data are collected into The word occurred in text information, so as to obtain the media text fragments comprising POI title.Can also use similarly Darts (even numbers group dictionary tree searching algorithm, Double-ARray Trie System) model.
2) previously according to the full dose POI name set and multiple media text fragments for having marked related POI, to condition Random field practice identification model is trained, and then practising identification model by trained condition random field can be to collection To network media text information judge whether it is the associated media text fragments of POI, compared to mode 1) can effectively avoid The case where semantic ambiguity, promotes accuracy of identification.
3) by generating accordingly the word and phrase that occur in multiple media text fragments for having marked related POI respectively Words feature vector goes out previously according to the full dose POI name set in multiple media text fragments for having marked related POI Existing words feature vector, is trained BI-LSTM-CRF model, can effectively improve entity recognition model to text information In sew the accuracy of identification of content before and after relevant to practical POI title text, to improved on the whole to the associated media of POI The accuracy of identification of text fragments.
And therefrom further analysis obtains media fragment relevant to POI trend of the times state, for judging the trend of the times shape of POI State, such as news item content are certain one floor Pizza Hut of mansion finishing, then the relevant target POI of this news data is exactly ground Location is located at the Pizza Hut of one floor of mansion, this news data obviously can be used as the associated network media data of target POI
S303 is obtained from media text fragments relevant to POI title and target POI according to the target POI title Relevant media text fragments.
Step S302 can filter out all and POI from the network media text information in the first time period being collected into Associated media text fragments, this step further therefrom obtain media text relevant to target POI according to the title of target POI This segment.
S304, in media text fragments relevant to target POI, to simultaneously comprising the target POI title and other The media text fragments of POI title carry out natural language processing, obtain the relations of dependence between the target POI and other POI.
I.e. there are two POI of the relations of dependence to occur firstly the need of in the same media text fragments for judgement, and then POI Trend of the times state obtaining apparatus carries out natural language processing to the media text fragments that two POI titles occur simultaneously, such as can be with Including semantic analysis and Keywords matching, such as is repeatedly occurring " dining room A in the mansion B " in news or " going to the mansion B 2 The text fragments of the similar contents such as the layer arrival dining room A ", can determine after semantic analysis is handled and exist between the two POI The relations of dependence.Carry out Keywords matching use keyword can such as " ... in ... ", " ... it is interior ", " ... it is inner ... " etc..
S305, according to there are the titles of other POI of the relations of dependence with the target POI, from the network being collected into Extract that there are the relevant media text fragments of other POI of the relations of dependence to the target POI in media text information.
S306, to the media text fragments relevant to target POI and with the target POI, there are the relations of dependence respectively The relevant media text fragments of other POI carry out natural language processing, obtain the target POI trend of the times data and with the mesh Mark POI there are the trend of the times data of other POI of the relations of dependence as with the associated network media data of target POI.
Natural language processing is carried out to the media text fragments relevant to target POI, such as may include semantic point Analysis and Keywords matching, can analyze the trend of the times data for obtaining target POI.Wherein reflect that the keyword of trend of the times state usually wraps Include: finishing, normally do business, close, stop doing business, brand upgrades, moves, studies in a school, by will include target POI title and it is above-mentioned instead The sentence of the keyword of gesture of appearing before one's eyes state carries out semantic analysis, can be obtained the trend of the times data of target POI.Similarly to the mesh Mark POI there are the relevant media text fragments of the relations of dependence other POI carry out natural language processing can analyze to obtain with it is described There are the trend of the times data of other POI of the relations of dependence by target POI.
The network information characteristic of target POI is converted corresponding trend of the times feature vector by S307.
Will include with the associated network media data of target POI in the present embodiment the target POI trend of the times data and There are the trend of the times data of other POI of the relations of dependence to be converted into corresponding trend of the times feature vector with the target POI.By network matchmaker The process that volume data is converted into corresponding trend of the times feature vector is considered as a kind of process of normalized.According to above-mentioned acquisition The combination of the trend of the times data of each different situations arrived or at least two trend of the times data can determine target POI different dimensions Trend of the times feature vector value, all dimensions of setting the trend of the times feature vector value composition target POI trend of the times feature vector, specifically In realization, each network can be determined by the feature vector value comparison table of preset trend of the times data and respective dimensions The corresponding feature vector value of information characteristics data.
The trend of the times feature vector of target POI is inputted trained trend of the times state judgment models, obtains target by S308 The trend of the times state of POI.
S309, by the trend of the times state output of target POI to application scenarios.
The application scenarios can such as Fig. 4 or electronic map application scenarios shown in fig. 5, can also be for example by trend of the times shape The changed POI of state is disclosed or is broadcasted, or is changed user's stroke planning according to the POI that trend of the times state change occurs or led Air route line etc..
The present embodiment analyzes network media data, obtain include target POI trend of the times data and with the target POI there are including the trend of the times data of other POI of the relations of dependence with target POI associated network media data, in conjunction with process Trained trend of the times state judgment models judge the trend of the times state of target POI, it is no longer necessary to inefficient hand data collection, The process for reporting and auditing, thus the POI Up-to-date state being effectively ensured in electronic map data.
Fig. 9 is referred to, for the embodiment of the invention provides a kind of structural schematic diagrams of POI trend of the times state obtaining apparatus.Such as Shown in Fig. 9, the POI trend of the times state obtaining apparatus of the embodiment of the present invention may include:
The network information obtains module 910, for obtaining and the associated network information characteristic of target POI.
The associated network information characteristic of described and target POI includes wireless access point corresponding with the target POI Access data, with the associated networks congestion control data of the target POI or with the associated network media number of the target POI According to.
Wherein, the network information, which obtains module 910 and obtains with the associated network media data of target POI, includes:
Collect the network media text information in first time period；
Matchmaker relevant to target POI is extracted from the network media text information being collected into according to target POI title Body text fragments are specifically as follows according to full dose POI name set and first instance identification model from the network being collected into Media text fragments relevant to POI title are extracted in media text information, the first instance identification model includes AC automatic Machine, condition random field entity recognition model or two-way shot and long term remember condition random field neural network model, and then according to described Target POI title obtains media text fragments relevant with target POI from media text fragments relevant to POI title, Described in first instance identification model can according to the full dose POI name set and it is multiple marked related POI media text This segment is trained as training sample；
Natural language processing is carried out to the media text fragments relevant to target POI, obtains showing for the target POI Gesture data as with the associated network media data of target POI.It can specifically include in media text piece relevant to target POI Duan Zhong carries out natural language processing to the media text fragments simultaneously comprising the target POI title and other POI titles, obtains The relations of dependence between the target POI and other POI, then according to the target POI there are the relations of dependence other The title of POI, from the network media text information being collected into extract with the target POI there are the relations of dependence other The relevant media text fragments of POI, so respectively to the media text fragments relevant to target POI and with the target There are the relevant media text fragments of the relations of dependence other POI to carry out natural language processing by POI, obtains showing for the target POI Gesture data and with the target POI there are the trend of the times data of other POI of the relations of dependence as with the associated network matchmaker of target POI Volume data.
In an alternative embodiment, the network information obtain module 910 can also according between target POI and other POI according to Attached relationship, using there are the network information characteristics of other POI of the relations of dependence as the associated net of target POI with target POI Network information characteristics data may further include and the target POI phase with the associated network information characteristic of target POI The network information characteristic of pass and there are the relevant network information characteristics of other POI of the relations of dependence to the target POI According to.
Trend of the times feature conversion module 920, it is special that the network information characteristic for will acquire is converted into the corresponding trend of the times Levy vector.
Trend of the times condition judgment module 930, the trend of the times feature vector for that will convert obtained target POI input trend of the times state Judgment models obtain the trend of the times state of target POI.
The trend of the times feature vector input trend of the times state judgement for the target POI that trend of the times condition judgment module 930 obtains conversion Model, so that it may which judgement obtains the trend of the times state of target POI.
In the specific implementation, trend of the times state judgment models can use GBDT (Gradient Boosting Decision Tree, gradient promote decision tree), XGboost machine learning function library or DNN (Deep Neural Networks, depth nerve Network) etc. machine learning models.
In an alternative embodiment, trend of the times state judgment models can judge that submodel merges by least two trend of the times states It arrives, wherein each trend of the times state judges that submodel is corresponding at least one trend of the times feature vector.
Trend of the times state output module 940, for exporting the trend of the times state of the target POI.
Specifically, trend of the times state output module 940 can be used for marking showing for the target POI in electronic map data Gesture state；Or for meeting the multiple of the POI retrieval request if inquiring when carrying out POI inquiry according to POI retrieval request Include the target POI in POI, is then carried in the POI query result of return in the multiple POI and exclude the target POI Other POI, or in the POI query result of return by the target POI in the multiple POI display sequence setting exist Finally.
In an alternative embodiment, POI trend of the times state obtaining apparatus further can also include:
Judgment models training module 950, for according to multiple POI for having marked trend of the times state and get with it is each The corresponding trend of the times feature vector of POI for having marked trend of the times state, is trained the trend of the times state judgment models.That is trend of the times shape State judgment models can be judgment models training module 950 according to multiple POI for having marked trend of the times state and get with What each corresponding trend of the times feature vector of POI for having marked trend of the times state was trained, multiple trend of the times states that marked The corresponding trend of the times state of POI is training sample.And the trend of the times feature corresponding with each POI for having marked trend of the times state to Amount can be converted by trend of the times feature conversion module 920 according to the network information characteristic of each POI for having marked trend of the times state It obtains.It is understood that by constantly trained trend of the times state judgment models, by the trend of the times feature vector of multiple sample POI The judging result for substituting into trend of the times state judgment models progress trend of the times state can constantly approach the mark knot of the multiple sample POI Fruit, the trend of the times state judgment models that judging result is finally most approached to sample POI annotation results are determined as the trend of the times that training obtains State judgment models.
If trend of the times state judgment models judge that submodel merges to obtain by least two trend of the times states, judgment models training mould Block 950 can be according to multiple POI (i.e. sample POI) for having marked trend of the times state and what is got each marked trend of the times state POI trend of the times feature vector, submodel, which is trained, to be judged to corresponding trend of the times state, it is to be understood that by continuous Trained trend of the times state judges submodel, by the trend of the times feature vector of multiple sample POI substitute into trend of the times state judge submodel into The judging result of row trend of the times state can constantly approach the annotation results of the multiple sample POI, finally most force judging result The trend of the times state of nearly sample POI annotation results judges that submodel is determined as the trend of the times state that training obtains and judges submodel；
Then judgment models training module 950 is according to multiple POI for having marked trend of the times state, having marked with each of getting The corresponding trend of the times feature vector of POI and trained at least two trends of the times state for infusing trend of the times state judge submodule Type judges that the fusion parameters of submodel are trained to each trend of the times state is corresponded in the trend of the times state judgment models.To Judge that submodel judges submodule with trained corresponding each trend of the times state according to trained at least two trends of the times state The fusion parameters of type obtain trained trend of the times state judgment models.
POI trend of the times state obtaining apparatus in the present embodiment according to the associated network information characteristic of target POI and shows Gesture state judgment models can be realized and quickly judge the trend of the times state of target POI, it is no longer necessary to inefficient hand data collection, The process for reporting and auditing, thus the POI Up-to-date state being effectively ensured in electronic map data.
The embodiment of the invention also provides a kind of computer storage medium, the computer storage medium can store more Item instruction, described instruction are suitable for being loaded by processor and being executed the method and step such as above-mentioned Fig. 1-embodiment illustrated in fig. 8, specifically hold Row process may refer to Fig. 1-embodiment illustrated in fig. 8 and illustrate, herein without repeating.
Referring to Figure 10, for the embodiment of the invention provides a kind of structural schematic diagrams of server.As shown in Figure 10, described Server 1000 may include: at least one processor 1001, such as CPU, at least one network interface 1004, user interface 1003, memory 1005, at least one communication bus 1002.Wherein, communication bus 1002 is for realizing between these components Connection communication.Network interface 1004 optionally may include standard wireline interface and wireless interface (such as WI-FI interface).Storage Device 1005 can be high speed RAM memory, be also possible to non-labile memory (non-volatile memory), such as At least one magnetic disk storage.Memory 1005 optionally can also be that at least one is located remotely from depositing for aforementioned processor 1001 Storage device.As shown in Figure 10, as may include that operating system, network are logical in a kind of memory 1005 of computer storage medium Believe that module and POI trend of the times state obtain program.
In the concrete realization, the server in the embodiment of the present invention can (one can be to mass data using Hadoop Carry out the software frame of distributed treatment) and the big datas processing skill such as Hive (Tool for Data Warehouse based on Hadoop) Art, be deployed in by hundreds physics unit at physical cluster on.Storage can be summarized using Hive database in the cluster to receive Then the various network information characteristics collected carry out feature using Mapreduce (a kind of distributed computing framework model) It extracts and preliminary Feature Engineering calculates, unstructured database (such as BDB, Berkeley DB database) is written after extracting It saves in case using.
In server 1000 shown in Fig. 10, network interface 1004 is mainly used for carrying out data communication with external equipment, Such as the trend of the times that POI is exported with the associated network information characteristic of target POI and to external equipment is obtained from external equipment State；And processor 1001 can be used for that the traffic information stored in memory 1005 is called to obtain application program, and specifically hold The following operation of row:
It obtains and the associated network information characteristic of target POI；
The network information characteristic that will acquire is converted into corresponding trend of the times feature vector；
The trend of the times feature vector for the target POI that conversion is obtained inputs trend of the times state judgment models, obtains showing for target POI Gesture state；
Export the trend of the times state of the target POI.
In one embodiment, described to include and the target POI phase with the associated network information characteristic of target POI The network information characteristic of pass and there are the relevant network information characteristics of other POI of the relations of dependence to the target POI According to.
In one embodiment, described to include and POI pairs of the target with the associated network information characteristic of target POI The access data for the wireless access point answered are closed with the associated networks congestion control data of the target POI or with the target POI The network media data of connection.
In one embodiment, it is specific with the associated network media data of target POI to execute acquisition for the processor 1001 Include:
Collect the network media text information in first time period；
Matchmaker relevant to target POI is extracted from the network media text information being collected into according to target POI title Body text fragments；
Natural language processing is carried out to the media text fragments relevant to target POI, obtains showing for the target POI Gesture data as with the associated network media data of target POI.
In one embodiment, the processor 1001 is executed according to target POI title from the network matchmaker being collected into Media text fragments relevant to target POI are extracted in body text information to specifically include:
According to full dose POI name set and first instance identification model from the network media text information being collected into Media text fragments relevant to POI title are extracted, the first instance identification model includes AC automatic machine, condition random field reality Body identification model or two-way shot and long term remember condition random field neural network model；
It is obtained from media text fragments relevant to POI title according to the target POI title relevant with target POI Media text fragments.
In one embodiment, the processor 1001 is being executed according to full dose POI name set and first instance identification Model is also held before extracting media text fragments relevant to POI title in the network media text information being collected into The following operation of row:
It is real to described first according to the full dose POI name set and multiple media text fragments for having marked related POI Body identification model is trained.
In one embodiment, the processor 1001 execute to the media text fragments relevant to target POI into Row natural language processing, the trend of the times data for obtaining the target POI are specifically wrapped as with the associated network media data of target POI It includes:
In media text fragments relevant to target POI, to simultaneously comprising the target POI title and other POI The media text fragments of title carry out natural language processing, obtain the relations of dependence between the target POI and other POI；
According to there are the titles of other POI of the relations of dependence with the target POI, from the network media text being collected into Extract that there are the relevant media text fragments of other POI of the relations of dependence to the target POI in this information；
Respectively to the media text fragments relevant to target POI and with the target POI there are the relations of dependence other The relevant media text fragments of POI carry out natural language processing, obtain the target POI trend of the times data and with the target POI there are the trend of the times data of other POI of the relations of dependence as with the associated network media data of target POI.
In one embodiment, the processor 1001 is in the trend of the times feature vector for executing the target POI for obtaining conversion Trend of the times state judgment models are inputted, before obtaining the trend of the times state of target POI, also execute following operation:
According to multiple POI for having marked trend of the times state and get corresponding with each POI for having marked trend of the times state Trend of the times feature vector, the trend of the times state judgment models are trained.
In one embodiment, the trend of the times state judgment models judge that submodel merges by least two trend of the times states It arrives, wherein each trend of the times state judges that submodel is corresponding at least one trend of the times feature vector.
And then the processor 1001 inputs trend of the times state in the trend of the times feature vector for executing the target POI for obtaining conversion Judgment models before obtaining the trend of the times state of target POI, also execute following operation:
According to the trend of the times of multiple POI for having marked trend of the times state and each POI for having marked trend of the times state got Feature vector judges that submodel is trained to corresponding trend of the times state；
And according to multiple POI for having marked trend of the times state, get it is corresponding with each POI for having marked trend of the times state Trend of the times feature vector and trained at least two trends of the times state judge submodel, judge mould to the trend of the times state Each trend of the times state is corresponded in type judges that the fusion parameters of submodel are trained.
In one embodiment, the trend of the times state that the processor 1001 executes the output target POI specifically includes:
The trend of the times state of the target POI is marked in electronic map data.
In another embodiment, the trend of the times state that the processor 1001 executes the output target POI specifically includes:
When carrying out POI inquiry according to POI retrieval request, if inquiring in the multiple POI for meeting the POI retrieval request Including the target POI, then other that the target POI is excluded in the multiple POI are carried in the POI query result of return POI, or in the POI query result of return by the target POI in the multiple POI display sequence be arranged last.
Server in the present embodiment can judge according to the associated network information characteristic of target POI and trend of the times state Model is realized and quickly judges the trend of the times state of target POI, it is no longer necessary to which inefficient hand data collection is reported and audited Process, thus the POI Up-to-date state being effectively ensured in electronic map data.
Those of ordinary skill in the art will appreciate that realizing all or part of the process in above-described embodiment method, being can be with Relevant hardware is instructed to complete by computer program, the program can be stored in a computer-readable storage medium In, the program is when being executed, it may include such as the process of the embodiment of above-mentioned each method.Wherein, the storage medium can be magnetic Dish, CD, read-only memory (Read-Only Memory, ROM) or random access memory (Random Access Memory, RAM) etc..
The above disclosure is only the preferred embodiments of the present invention, cannot limit the right model of the present invention with this certainly It encloses, therefore equivalent changes made in accordance with the claims of the present invention, is still within the scope of the present invention.

Claims (15)

1. a kind of POI trend of the times state acquiring method, which is characterized in that the described method includes:
It obtains and the associated network information characteristic of target POI；
The network information characteristic that will acquire is converted into corresponding trend of the times feature vector；
The trend of the times feature vector for the target POI that conversion is obtained inputs trend of the times state judgment models, obtains the trend of the times shape of target POI State；
Export the trend of the times state of the target POI.
2. method as shown in claim 1, which is characterized in that the described and associated network information characteristic packet of target POI It includes network information characteristic relevant to the target POI and there are other POI of the relations of dependence are related with the target POI Network information characteristic.
3. the method as described in claim 1, which is characterized in that the described and associated network information characteristic packet of target POI Include the access data and the associated networks congestion control data of the target POI of wireless access point corresponding with the target POI Or with the associated network media data of target POI.
4. method as claimed in claim 3, which is characterized in that the acquisition and the associated network media data packet of target POI It includes:
Collect the network media text information in first time period；
Media text relevant to target POI is extracted from the network media text information being collected into according to target POI title This segment；
Natural language processing is carried out to the media text fragments relevant to target POI, obtains the trend of the times number of the target POI According to as with the associated network media data of target POI.
5. method as claimed in claim 4, which is characterized in that it is described according to target POI title from the network being collected into Media text fragments relevant to target POI are extracted in media text information includes:
It is extracted from the network media text information being collected into according to full dose POI name set and first instance identification model Media text fragments relevant to POI title, the first instance identification model include AC automatic machine, the knowledge of condition random field entity Other model or two-way shot and long term remember condition random field neural network model；
Media relevant with target POI are obtained from media text fragments relevant to POI title according to the target POI title Text fragments.
6. method as claimed in claim 5, which is characterized in that described to be identified according to full dose POI name set and first instance Model also wraps before extracting media text fragments relevant to POI title in the network media text information being collected into It includes:
According to the full dose POI name set and multiple media text fragments for having marked related POI, the first instance is known Other model is trained.
7. method as claimed in claim 4, which is characterized in that described to the media text fragments relevant to target POI Carry out natural language processing, obtain the trend of the times data of the target POI as with the associated network media data packet of target POI It includes:
In media text fragments relevant to target POI, to simultaneously comprising the target POI title and other POI titles Media text fragments carry out natural language processing, obtain the relations of dependence between the target POI and other POI；
According to there are the titles of other POI of the relations of dependence with the target POI, from the network media text envelope being collected into Extract that there are the relevant media text fragments of other POI of the relations of dependence to the target POI in breath；
To the media text fragments relevant to target POI and with the target POI, there are other POI phases of the relations of dependence respectively The media text fragments of pass carry out natural language processing, obtain the trend of the times data of the target POI and exist with the target POI The trend of the times data of other POI of the relations of dependence as with the associated network media data of target POI.
8. the method as described in right wants 1, which is characterized in that the trend of the times feature vector of the target POI for obtaining conversion is defeated Enter trend of the times state judgment models, before obtaining the trend of the times state of target POI further include:
According to multiple POI for having marked trend of the times state and get corresponding existing with each POI for having marked trend of the times state Gesture feature vector is trained the trend of the times state judgment models.
9. the method as described in claim 1, which is characterized in that the trend of the times state judgment models are by least two trend of the times states Judge that submodel merges to obtain, wherein each trend of the times state judges that submodel is corresponding at least one trend of the times feature vector.
10. method as claimed in claim 9, which is characterized in that the trend of the times feature vector of the target POI for obtaining conversion Trend of the times state judgment models are inputted, before obtaining the trend of the times state of target POI further include:
According to the trend of the times feature of multiple POI for having marked trend of the times state and each POI for having marked trend of the times state got Vector judges that submodel is trained to corresponding trend of the times state；
And according to multiple POI for having marked trend of the times state, the trend of the times corresponding with each POI for having marked trend of the times state got Feature vector and trained at least two trends of the times state judge submodel, in the trend of the times state judgment models Corresponding each trend of the times state judges that the fusion parameters of submodel are trained.
11. such as method of any of claims 1-10, which is characterized in that the trend of the times of the output target POI State includes:
The trend of the times state of the target POI is marked in electronic map data.
12. such as method of any of claims 1-10, which is characterized in that the trend of the times of the output target POI State includes:
When carrying out POI inquiry according to POI retrieval request, include if inquiring in the multiple POI for meeting the POI retrieval request The target POI then carries other POI that the target POI is excluded in the multiple POI in the POI query result of return, Or the display sequence in the POI query result of return by the target POI in the multiple POI is arranged last.
13. a kind of POI trend of the times state obtaining apparatus, which is characterized in that the equipment includes:
The network information obtains module, for obtaining and the associated network information characteristic of target POI；
Trend of the times feature conversion module, the network information characteristic for will acquire are converted into corresponding trend of the times feature vector；
Trend of the times condition judgment module, the trend of the times feature vector for that will convert obtained target POI input trend of the times state and judge mould Type obtains the trend of the times state of target POI；
Trend of the times state output module, for exporting the trend of the times state of the target POI.
14. a kind of computer storage medium, which is characterized in that the computer storage medium is stored with a plurality of instruction, the finger It enables and is suitable for being loaded by processor and executing following steps:
It obtains and the associated network information characteristic of target POI；
The network information characteristic that will acquire is converted into corresponding trend of the times feature vector；
The trend of the times feature vector for the target POI that conversion is obtained inputs trend of the times state judgment models, obtains the trend of the times shape of target POI State；
Export the trend of the times state of the target POI.
15. a kind of server characterized by comprising processor and memory；Wherein, the memory is stored with computer Program, the computer program are suitable for being loaded by the processor and executing following steps:
It obtains and the associated network information characteristic of target POI；
The network information characteristic that will acquire is converted into corresponding trend of the times feature vector；
The trend of the times feature vector for the target POI that conversion is obtained inputs trend of the times state judgment models, obtains the trend of the times shape of target POI State；
Export the trend of the times state of the target POI.
CN201711260201.5A 2017-12-04 2017-12-04 POI (Point of interest) situation acquisition method and equipment, storage medium and server thereof Active CN110019616B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201711260201.5A CN110019616B (en) 2017-12-04 2017-12-04 POI (Point of interest) situation acquisition method and equipment, storage medium and server thereof

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201711260201.5A CN110019616B (en) 2017-12-04 2017-12-04 POI (Point of interest) situation acquisition method and equipment, storage medium and server thereof

Publications (2)

Publication Number Publication Date
CN110019616A true CN110019616A (en) 2019-07-16
CN110019616B CN110019616B (en) 2022-07-12

Family

ID=67186530

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201711260201.5A Active CN110019616B (en) 2017-12-04 2017-12-04 POI (Point of interest) situation acquisition method and equipment, storage medium and server thereof

Country Status (1)

Country Link
CN (1) CN110019616B (en)

Cited By (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110347775A (en) * 2019-07-17 2019-10-18 北京百度网讯科技有限公司 Point of interest state correction method, apparatus, equipment and computer readable storage medium
CN110851738A (en) * 2019-10-28 2020-02-28 百度在线网络技术（北京）有限公司 Method, device and equipment for acquiring POI state information and computer storage medium
CN110990728A (en) * 2019-12-03 2020-04-10 汉海信息技术(上海)有限公司 Method, device and equipment for managing point of interest information and storage medium
CN111198916A (en) * 2020-01-03 2020-05-26 北京明略软件系统有限公司 Data transmission method and device, electronic equipment and storage medium
CN111782973A (en) * 2020-06-04 2020-10-16 汉海信息技术(上海)有限公司 Interest point state prediction method and device, electronic equipment and storage medium
CN111832483A (en) * 2020-07-14 2020-10-27 北京百度网讯科技有限公司 Method, device, equipment and storage medium for identifying validity of interest point
CN111860503A (en) * 2020-07-16 2020-10-30 北京奇虎科技有限公司 Information point validity identification method, device, equipment and storage medium
CN111858787A (en) * 2019-09-24 2020-10-30 北京嘀嘀无限科技发展有限公司 POI information acquisition method and device
CN112330379A (en) * 2020-11-25 2021-02-05 税友软件集团股份有限公司 Invoice content generation method and system, electronic equipment and storage medium
CN113222492A (en) * 2021-03-29 2021-08-06 北京中交兴路信息科技有限公司 Method and device for judging vehicle driving line type, storage medium and terminal
CN114860836A (en) * 2022-05-24 2022-08-05 北京百度网讯科技有限公司 Method, device, equipment and medium for mining failure interest points

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102788586A (en) * 2011-05-20 2012-11-21 广东瑞图万方科技股份有限公司 Method and system for incremental updating of navigation electronic map
CN103064953A (en) * 2012-12-28 2013-04-24 武汉大学 Point of interest (POI) updating method based on location check-in data
CN104866542A (en) * 2015-05-05 2015-08-26 腾讯科技（深圳）有限公司 POI data verification method and device
US20170180928A1 (en) * 2015-12-21 2017-06-22 Amazon Technologies, Inc. Maintaining privacy in location-based operations
CN107133900A (en) * 2017-04-05 2017-09-05 中国农业大学 Urban land mixing utilizes feature grid computational methods and device

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102788586A (en) * 2011-05-20 2012-11-21 广东瑞图万方科技股份有限公司 Method and system for incremental updating of navigation electronic map
CN103064953A (en) * 2012-12-28 2013-04-24 武汉大学 Point of interest (POI) updating method based on location check-in data
CN104866542A (en) * 2015-05-05 2015-08-26 腾讯科技（深圳）有限公司 POI data verification method and device
US20170180928A1 (en) * 2015-12-21 2017-06-22 Amazon Technologies, Inc. Maintaining privacy in location-based operations
CN107133900A (en) * 2017-04-05 2017-09-05 中国农业大学 Urban land mixing utilizes feature grid computational methods and device

Cited By (22)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110347775A (en) * 2019-07-17 2019-10-18 北京百度网讯科技有限公司 Point of interest state correction method, apparatus, equipment and computer readable storage medium
CN110347775B (en) * 2019-07-17 2022-11-08 北京百度网讯科技有限公司 Interest point state correction method, device, equipment and computer readable storage medium
CN111858787A (en) * 2019-09-24 2020-10-30 北京嘀嘀无限科技发展有限公司 POI information acquisition method and device
WO2021082397A1 (en) * 2019-10-28 2021-05-06 百度在线网络技术（北京）有限公司 Method and device for acquiring poi state information, apparatus, and computer storage medium
US11709999B2 (en) 2019-10-28 2023-07-25 Baidu Online Network Technology (Beijing) Co., Ltd. Method and apparatus for acquiring POI state information, device and computer storage medium
CN110851738A (en) * 2019-10-28 2020-02-28 百度在线网络技术（北京）有限公司 Method, device and equipment for acquiring POI state information and computer storage medium
JP2022511593A (en) * 2019-10-28 2022-02-01 バイドゥ オンライン ネットワーク テクノロジー（ペキン） カンパニー リミテッド Methods, devices, devices, programs and computer storage media for acquiring POI status information
JP7214949B2 (en) 2019-10-28 2023-01-31 バイドゥ オンライン ネットワーク テクノロジー（ペキン） カンパニー リミテッド METHOD, APPARATUS, DEVICE, PROGRAM AND COMPUTER STORAGE MEDIA FOR ACQUIRING POI STATE INFORMATION
CN110990728B (en) * 2019-12-03 2023-09-12 汉海信息技术(上海)有限公司 Method, device, equipment and storage medium for managing interest point information
CN110990728A (en) * 2019-12-03 2020-04-10 汉海信息技术(上海)有限公司 Method, device and equipment for managing point of interest information and storage medium
CN111198916A (en) * 2020-01-03 2020-05-26 北京明略软件系统有限公司 Data transmission method and device, electronic equipment and storage medium
CN111198916B (en) * 2020-01-03 2023-12-08 北京明略软件系统有限公司 Data transmission method and device, electronic equipment and storage medium
CN111782973A (en) * 2020-06-04 2020-10-16 汉海信息技术(上海)有限公司 Interest point state prediction method and device, electronic equipment and storage medium
CN111832483B (en) * 2020-07-14 2024-03-08 北京百度网讯科技有限公司 Point-of-interest validity identification method, device, equipment and storage medium
CN111832483A (en) * 2020-07-14 2020-10-27 北京百度网讯科技有限公司 Method, device, equipment and storage medium for identifying validity of interest point
CN111860503A (en) * 2020-07-16 2020-10-30 北京奇虎科技有限公司 Information point validity identification method, device, equipment and storage medium
CN112330379B (en) * 2020-11-25 2023-10-31 税友软件集团股份有限公司 Invoice content generation method, invoice content generation system, electronic equipment and storage medium
CN112330379A (en) * 2020-11-25 2021-02-05 税友软件集团股份有限公司 Invoice content generation method and system, electronic equipment and storage medium
CN113222492A (en) * 2021-03-29 2021-08-06 北京中交兴路信息科技有限公司 Method and device for judging vehicle driving line type, storage medium and terminal
CN113222492B (en) * 2021-03-29 2024-05-03 北京中交兴路信息科技有限公司 Method and device for discriminating type of vehicle driving line, storage medium and terminal
CN114860836B (en) * 2022-05-24 2023-03-10 北京百度网讯科技有限公司 Method, device, equipment and medium for mining failure interest points
CN114860836A (en) * 2022-05-24 2022-08-05 北京百度网讯科技有限公司 Method, device, equipment and medium for mining failure interest points

Also Published As

Publication number Publication date
CN110019616B (en) 2022-07-12

Similar Documents

Publication Publication Date Title
CN110019616A (en) A kind of POI trend of the times state acquiring method and its equipment, storage medium, server
JP2021108183A (en) Method, apparatus, device and storage medium for intention recommendation
CN110597981B (en) Network news summary system for automatically generating summary by adopting multiple strategies
US9514425B2 (en) Method and system for providing user-customized contents
CN105843841A (en) Small file storing method and system
CN110825956A (en) Information flow recommendation method and device, computer equipment and storage medium
CN113254630B (en) Domain knowledge map recommendation method for global comprehensive observation results
CN111159341B (en) Information recommendation method and device based on user investment and financial management preference
CN107092639A (en) A kind of search engine system
CN111708774B (en) Industry analytic system based on big data
CN108062366B (en) Public culture information recommendation system
CN107918657A (en) The matching process and device of a kind of data source
WO2023108980A1 (en) Information push method and device based on text adversarial sample
CN102222098A (en) Method and system for pre-fetching webpage
CN110795613B (en) Commodity searching method, device and system and electronic equipment
CN110362663A (en) Adaptive more perception similarity detections and parsing
CN115238173B (en) Behavior analysis and medical service pushing method, equipment and medium based on big data
CN110134845A (en) Project public sentiment monitoring method, device, computer equipment and storage medium
CN109739965A (en) Moving method and device, equipment, the readable storage medium storing program for executing of cross-cutting dialog strategy
CN112015908A (en) Knowledge graph construction method and system, and query method and system
CN110069686A (en) User behavior analysis method, apparatus, computer installation and storage medium
CN111259975B (en) Method and device for generating classifier and method and device for classifying text
CN111753151A (en) Service recommendation method based on internet user behaviors
CN103365645B (en) Method and equipment for maintaining a software system
Mohapatra et al. Query answering for kisan call centerwith lda/lsi

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant