CN109857859A - Processing method, device, equipment and the storage medium of news information - Google Patents
Processing method, device, equipment and the storage medium of news information Download PDFInfo
- Publication number
- CN109857859A CN109857859A CN201811581267.9A CN201811581267A CN109857859A CN 109857859 A CN109857859 A CN 109857859A CN 201811581267 A CN201811581267 A CN 201811581267A CN 109857859 A CN109857859 A CN 109857859A
- Authority
- CN
- China
- Prior art keywords
- news
- event
- processed
- similar
- preset
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Landscapes
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The application provides processing method, device, equipment and the storage medium of a kind of news information, wherein, this method comprises: from preset event base, determine at least one similar news similar to news to be processed, it include at least one event cluster corresponding with event in event base, there is at least one news, each similar news is belonging respectively to different events in each event cluster；By news to be processed, news similar with each is combined respectively, constitutes the first different news pair；By each first news to being input in preset two disaggregated model, event corresponding to news to be processed is obtained, two disaggregated models are N number of second news using the event that is labelled with to being trained.Using there is the model of supervision to the first news to identifying, event corresponding to the news to be processed of the first news centering is accurately determined；What can be directed to classifies to news, improves the verification and measurement ratio for belonging to the news of the same event.
Description
Technical field
The invention relates to knowledge information processing technical field more particularly to a kind of processing method of news information,
Device, equipment and storage medium.
Background technique
In the information age, the network information is in explosive growth.Various information can be presented to the user now；When user thinks
It is to be understood that when some specific event, such as when understanding the thing occurred in the recent period, concern someone, user need oneself from big
In news of the amount without screening arrangement, important information is picked out, wherein it include the news of non-event in each news,
Non-event for example has advertisement, health consultation etc..The new of the event of oneself concern can be quickly viewed for the ease of user
It hears, can classify to news, be then presented to the user the news only comprising user's event of interest.
In the prior art, news can be clustered using clustering method, it is specifically, right according to the topic of news
News carries out clustering processing, obtains multiple clusters, a corresponding event in each cluster, same including belonging in each cluster
The news of event obtains the news of user's event of interest to get to belonging to the news of different event.
However in the prior art, there is redundancy in the topic of news, and then news is carried out according to the topic of news
After clustering processing, the obtained news in cluster is simultaneously impure, i.e., the news in cluster is simultaneously not belonging to the same event, can not be accurate
Determine event corresponding to news；Also, clustering method belongs to a kind of unsupervised approaches, and unsupervised approaches in corpus using depositing
Bulk redundancy information do cluster calculation, and then further lead to the news in cluster and be not belonging to the same event.To
In the prior art can not accurately to news carry out classification processing, for belong to the same event news verification and measurement ratio compared with
It is low.
Summary of the invention
The embodiment of the present application provides processing method, device, equipment and the storage medium of a kind of news information, for solving
Classification processing accurately can not be carried out to news in above scheme, it is lower for belonging to the verification and measurement ratio of news of the same event
The problem of.
The application first aspect provides a kind of processing method of news information, comprising:
Obtain news to be processed；
From preset event base, at least one similar news similar to the news to be processed is determined, wherein institute
Stating in event base includes at least one event cluster, each described event cluster is corresponding with a kind of event, each described event cluster
In there is at least one news, the similar news of each of at least one described similar news is belonging respectively to different events；
It combines the news to be processed with similar news described in each respectively, constitutes the first different news pair,
Wherein, the first news centering includes the news to be processed and the similar news；
By each first news to being input in preset two disaggregated model, obtain corresponding to the news to be processed
Event, wherein two disaggregated model is M the using being labelled with N number of second news of event to being trained
Two news to each of the second news centering include belonging to two news of same event, N-M the second news centerings
Each second news centering include belonging to two news of different event, N, M are positive integer more than or equal to 1, and N is greater than
M。
Further, each first news is obtained described to be processed to being input in preset two disaggregated model
Event corresponding to news, comprising:
By each first news to being input in preset two disaggregated model, each described first news is exported to category
In the probability value of similar events；
If it is determined that maximum probability value is greater than preset threshold, it is determined that the similar news of the maximum news centering of probability value is returned
The event of category is event corresponding to the news to be processed.
Further, the method, further includes:
If it is determined that maximum probability value is less than or equal to preset threshold, then a new thing is created for the news to be processed
Part, and by the new events as event corresponding to the news to be processed.
Further, by each first news to being input in preset two disaggregated model, export each described
One news is to the probability value for belonging to similar events, comprising:
By each first news to being input in preset two disaggregated model, each described first news is exported to category
In the probability value for the event that the similar news of each described news centering is belonged to.
It is further, described to obtain news to be processed, comprising:
Obtain news agregator to be processed, wherein include at least one news in the news agregator；
The news agregator is input in two disaggregated model, to filter out the news of non-event, after obtaining processing
News agregator, wherein include the news to be processed in treated the news agregator.
Further, the method, further includes:
Obtain the M the second news to the N-M the second news pair；
, to feature calculation is carried out, at least one feature is obtained to the N-M the second news to the M the second news
Information；
The M the second news are input to, the N-M the second news pair and at least one characteristic information
In two disaggregated models to be trained, preset two disaggregated model is obtained.
Further, the M the second news pair are obtained, comprising:
Multiple news to be processed are clustered, the event base is obtained；
Receive markup information, wherein the markup information includes to new in each of described event base event cluster
The update information of news, and according to the markup information, whether the news in each of described event base event cluster is belonged to
Current event cluster is modified, and obtains revised event base；
The M the second news pair are obtained from the revised event base.
Further, the N-M the second news pair are obtained, comprising:
According to preset keyword, multiple news relevant to the keyword are inquired from preset database；
Will multiple news relevant to the keyword, form the N-M the second news pair.
Further, will multiple news relevant to the keyword, form the N-M the second news pair, comprising:
According to the descending of the matching degree with keyword, place is ranked up to multiple news relevant to the keyword
Reason, multiple news after being sorted；
The news for belonging to different event in multiple news after choosing the sequence, the multiple news selected；
By the multiple news selected, the N-M the second news pair are formed.
Further, it, to being input in preset two disaggregated model, is obtained described to be processed by each first news
News corresponding to after event, further includes:
By the news to be processed, it is right to be put into the institute of event corresponding to news to be processed described in the event base
In the event cluster answered.
The application second aspect provides a kind of processing unit of news information, comprising:
First acquisition unit, for obtaining news to be processed；
First determination unit, for determining similar with the news to be processed at least one from preset event base
A similar news, wherein include at least one event cluster, each described event cluster and a kind of event pair in the event base
It answers, there is at least one news, the similar news of each of at least one described similar news in each described event cluster
It is belonging respectively to different events；
First processing units are constituted for combining the news to be processed with similar news described in each respectively
The first different news pair, wherein the first news centering includes the news to be processed and the similar news；
Second determination unit, it is described for being input in preset two disaggregated model, obtaining each first news
Event corresponding to news to be processed, wherein two disaggregated model be using be labelled with N number of second news of event into
Row training obtains, M the second news to each of the second news centering include belong to two news of same event,
N-M the second news to each of the second news centering include belonging to two news of different event, N, M be greater than etc.
In 1 positive integer, N is greater than M.
Further, second determination unit, comprising:
Output module, for each first news to being input in preset two disaggregated model, to be exported each institute
The first news is stated to the probability value for belonging to similar events；
First determining module is used for if it is determined that maximum probability value is greater than preset threshold, it is determined that probability value is maximum new
The event that the similar news of centering is belonged to is heard, is event corresponding to the news to be processed.
Further, second determination unit, further includes:
Second determining module, for if it is determined that maximum probability value is then described to be processed less than or equal to preset threshold
News creates a new events, and by the new events as event corresponding to the news to be processed.
Further, the output module, is specifically used for:
By each first news to being input in preset two disaggregated model, each described first news is exported to category
In the probability value for the event that the similar news of each described news centering is belonged to.
Further, the first acquisition unit, comprising:
First obtains module, for obtaining news agregator to be processed, wherein includes at least one in the news agregator
News；
Filtering module, for the news agregator to be input in two disaggregated model, to filter out the new of non-event
It hears, the news agregator that obtains that treated, wherein include the news to be processed in treated the news agregator.
Further, described device, further includes:
Second acquisition unit, for obtaining the M the second news pair；
Third acquiring unit, for the N-M the second news pair；
Computing unit, for, to feature calculation is carried out, being obtained to the M the second news to the N-M the second news
To at least one characteristic information；
Output unit, for by the M the second news to, the N-M the second news pair and at least one
Characteristic information is input in two disaggregated models to be trained, and obtains preset two disaggregated model.
Further, the second acquisition unit, comprising:
Cluster module obtains the event base for clustering to multiple news to be processed；
Receiving module, for receiving markup information, wherein the markup information includes to each of described event base
The update information of news in event cluster, and according to the markup information, in each of described event base event cluster
Whether news, which belongs to current event cluster, is modified, and obtains revised event base；
Second obtains module, for obtaining the M the second news pair from the revised event base.
Further, the third acquiring unit, comprising:
Enquiry module, for being inquired from preset database related to the keyword according to preset keyword
Multiple news；
Comprising modules, for will multiple news relevant to the keyword, form the N-M the second news pair.
Further, the comprising modules, are specifically used for:
According to the descending of the matching degree with keyword, place is ranked up to multiple news relevant to the keyword
Reason, multiple news after being sorted；
The news for belonging to different event in multiple news after choosing the sequence, the multiple news selected；
By the multiple news selected, the N-M the second news pair are formed.
Further, described device, further includes:
The second processing unit, in second determination unit by each first news to being input to preset two points
In class model, after obtaining event corresponding to the news to be processed, by the news to be processed, it is put into the thing
In event cluster corresponding to event corresponding to news to be processed described in part library.
The application third aspect provides a kind of control equipment, comprising: transmitter, receiver, memory and processor；
The memory is for storing computer instruction；The processor by run memory storage it is described based on
The method that any implementation of first aspect provides is realized in the instruction of calculation machine.
The application fourth aspect provides a kind of storage medium, comprising: readable storage medium storing program for executing and computer instruction, the calculating
Machine instruction is stored in the readable storage medium storing program for executing；The computer instruction provides for realizing any implementation of first aspect
Method.
Processing method, device, equipment and the storage medium of news information provided by the embodiments of the present application, by obtain to
The news of processing；From preset event base, at least one similar news similar to news to be processed is determined, wherein thing
It include at least one event cluster in part library, each event cluster is corresponding with a kind of event, has at least one in each event cluster
A news, the similar news of each of at least one similar news are belonging respectively to different events；By news to be processed point
News not similar with each combines, and constitutes the first different news pair, wherein the first news centering includes news to be processed
The similar news with one；By each first news to being input in preset two disaggregated model, obtain corresponding to news to be processed
Event, wherein two disaggregated models are using being labelled with N number of second news of event to being trained.Due to two points
Class model is the second news using the event that is labelled with to being trained, so that two disaggregated models are that one kind has supervision
Model；Using there is the model of supervision to the first news to identifying, so accurately determine the first news centering to
Event corresponding to the news of processing；What can be directed to classifies to news, improves the news for belonging to the same event
Verification and measurement ratio.
Detailed description of the invention
In order to illustrate the technical solutions in the embodiments of the present application or in the prior art more clearly, 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 this Shen
Some embodiments please for those of ordinary skill in the art without any creative labor, can be with
Other attached drawings are obtained according to these attached drawings.
Fig. 1 is a kind of flow chart of the processing method of news information provided by the embodiments of the present application；
Fig. 2 is the flow chart of the processing method of another news information provided by the embodiments of the present application；
Fig. 3 is a kind of structural schematic diagram of the processing unit of news information provided by the embodiments of the present application；
Fig. 4 is the structural schematic diagram of the processing unit of another news information provided by the embodiments of the present application；
Fig. 5 is a kind of structural schematic diagram for controlling equipment provided by the embodiments of the present application.
Specific embodiment
To keep the purposes, technical schemes and advantages of the embodiment of the present application clearer, below in conjunction with the embodiment of the present application
In attached drawing, the technical scheme in the embodiment of the application is clearly and completely described, it is clear that described embodiment is
Some embodiments of the present application, instead of all the embodiments.Based on the embodiment in the application, those of ordinary skill in the art
All other embodiment obtained without creative efforts, shall fall in the protection scope of this application.
The term that presently filed embodiment part uses is only used for explaining the specific embodiment of the application, rather than
It is intended to limit the application.The part term in the application is explained below, in order to those skilled in the art understand that.
1) the corresponding event cluster (Cluster) of event (Event): event；News under one event is ownership
In same event.
2) news: also become newsletter archive, Domestic News, News Resources；In news have text information, pictorial information,
Video information, etc..
4) " multiple " refer to two or more, and other quantifiers are similar therewith."and/or" describes the pass of affiliated partner
Connection relationship indicates may exist three kinds of relationships, for example, A and/or B, can indicate: individualism A exists simultaneously A and B, individually
There are these three situations of B.Character "/" typicallys represent the relationship that forward-backward correlation object is a kind of "or".
5) " correspondence " also refers to a kind of incidence relation or binding relationship, and A is corresponding with B to refer to it being one between A and B
Kind incidence relation or binding relationship.
It should be pointed out that noun involved in the embodiment of the present application or term can be referred to mutually, repeat no more.
In the prior art, news can be clustered using clustering method, it is specifically, right according to the topic of news
News carries out clustering processing, obtains multiple clusters, a corresponding event in each cluster, same including belonging in each cluster
The news of event obtains the news of user's event of interest to get to belonging to the news of different event.
However in the prior art, there is redundancy in the topic of news, and then news is carried out according to the topic of news
After clustering processing, the obtained news in cluster is simultaneously impure, i.e., the news in cluster is simultaneously not belonging to the same event, can not be accurate
Determine event corresponding to news；Also, clustering method belongs to a kind of unsupervised approaches, and unsupervised approaches in corpus using depositing
Bulk redundancy information do cluster calculation, and then further lead to the news in cluster and be not belonging to the same event.To
In the prior art can not accurately to news carry out classification processing, for belong to the same event news verification and measurement ratio compared with
It is low, accurately the news only comprising user's event of interest can not be presented to the user.
In view of the above problems, the application proposes processing method, device, equipment and the storage of a kind of news information
Medium, can be using there is the model of supervision to the first news to identifying, and then accurately determines the first news centering
Event corresponding to news to be processed；What can be directed to classifies to news, improves the news for belonging to the same event
Verification and measurement ratio.The program is described in detail below by several specific embodiments.
Fig. 1 is a kind of flow chart of the processing method of news information provided by the embodiments of the present application, as shown in Figure 1, the party
Method, comprising:
Step 101 obtains news to be processed.
In this step, specifically, the executing subject of the embodiment of the present application can be terminal or server or new
Hear information processing unit or equipment or other can execute the device or equipment of the embodiment of the present application, the application does not limit
System.
News agregator to be processed is obtained first, wherein includes that at least one is to be processed in news agregator to be processed
News.For each news to be processed, the step 102-104 of the embodiment of the present application is executed.
For example, a news agregator to be processed is obtained, includes news to be processed in news agregator to be processed
A, news B to be processed, news C to be processed, news D to be processed, news E to be processed, news F to be processed.
Step 102, from preset event base, determine similar to news to be processed at least one similar news,
In, it include at least one event cluster in event base, each event cluster is corresponding with a kind of event, has extremely in each event cluster
Few news, the similar news of each of at least one similar news are belonging respectively to different events.
In this step, specifically, pre-setting an event base, there is Q event cluster in the event base, each
There is M news, Q, M are positive integer in event cluster；Each event cluster is corresponding with a kind of event, i.e., under each event cluster
News belong to same event.
Then, P similar news similar to news to be processed is obtained from the event base, wherein P is positive integer.
This step includes following several implementations.
The first implementation: obtaining the keyword of news to be processed, and the news in event base also has keyword；It can
To be matched and be analyzed according to the keyword of the news in the keyword and event base of news to be processed, so determine with
The similar news of news to be processed；By news similar with news to be processed, referred to as similar news；Then it weeds out similar
Belong to the news of event cluster in news, so that each similar news similar to news to be processed got respectively corresponds
Different event clusters, that is, each similar news similar from news to be processed got respectively correspond different events.
For example, for news A to be processed, similar news 1 similar to news A, phase are got from event base
Like news 2, similar news 3 and similar news 4, wherein similar news 1 belongs to event cluster a, and similar news 2 belongs to event cluster b,
Similar news 3 belongs to event cluster c, and similar news 4 belongs to event cluster d.
Second of implementation: according to event base, similar with news to be processed one under each event cluster is obtained
A similar news.
For example, for news A to be processed, event base have event cluster a, event cluster b, event cluster c, event cluster d,
Event cluster d, event cluster e；To determine one and the most similar similar news of news A in each event cluster, and then obtain
The similar news 2 under similar news 1, event cluster b under to event cluster a, the similar news 3 under event cluster c, under event cluster d
Similar news 5 under similar news 4, event cluster e.
Step 103, by news to be processed, news similar with each is combined respectively, constitutes the first different news pair,
Wherein, the first news centering includes news and a similar news to be processed.
In this step, specifically, news to be processed is matched two-by-two to P similar news, P first is obtained
News pair.P the first news to each of the first news centering include news and a similar news to be processed.
For example, for news A to be processed, similar news 1 under event cluster a, similar new under event cluster b is obtained
Hear 2, the similar news 3 under event cluster c, the similar news 4 under event cluster d, the similar news 5 under event cluster e；By news A and
Similar news 1 forms the first news to a, and news A and similar news 2 are formed the first news to b, by news A and similar news 3
The first news is formed to c, news A and similar news 4 are formed into the first news to d, by news A and the composition of similar news 5 first
News is to e.
Step 104, by each first news to being input in preset two disaggregated model, obtain corresponding to news to be processed
Event, wherein two disaggregated models are using being labelled with N number of second news of event to being trained, and M second is new
Hear to each of the second news centering include belonging to two news of same event, N-M the second news centering it is every
One the second news centering includes belonging to two news of different event, and N, M are the positive integer more than or equal to 1, and N is greater than M.
In this step, specifically, pre-setting two disaggregated models, which is to be got over using mark
N number of second news of part is to being trained.Wherein, part the second news centering of N number of second news centering is each
A second news to including belonging to two news of same event, residue the second news centering of N number of second news centering it is every
One the second news is to two news including belonging to different event.
For example, 6 the second news pair, respectively the second news are to a, the second news to b, the second news to c, second
News is to d, the second news to e, the second news to f；Second news belongs to event a to two news that a includes, and second is new
News belongs to event b to two news that b includes, and the second news belongs to event c to two news that c includes；Second
News belongs to different event to two news that d includes, and the second news is to belong to different event to e two news for including
, the second news belongs to different event to two news that f includes.
By each first news to being input in above-mentioned two disaggregated model, so that it may export thing corresponding to news to be processed
Part is any event.
The embodiment of the present application, by obtaining news to be processed；From preset event base, determining and to be processed news
At least one similar similar news, wherein include at least one event cluster, each event cluster and a kind of event in event base
It is corresponding, there is at least one news, the similar news of each of at least one similar news belongs to respectively in each event cluster
In different events；By news to be processed, news similar with each is combined respectively, constitutes the first different news pair,
In, the first news centering includes news and a similar news to be processed；By each first news to being input to preset two points
In class model, event corresponding to news to be processed is obtained, wherein two disaggregated models are N number of the using the event that is labelled with
Two news are to being trained.Since two disaggregated models are using being labelled with the second news of event to being trained
It arrives, so that two disaggregated models are a kind of models for having supervision；Using there is the model of supervision to the first news to identifying, into
And accurately determine event corresponding to the news to be processed of the first news centering；What can be directed to divides news
Class improves the verification and measurement ratio for belonging to the news of the same event.
Fig. 2 is the flow chart of the processing method of another news information provided by the embodiments of the present application, as shown in Fig. 2, should
Method, comprising:
Step 201, obtain M the second news to and N-M the second news pair, wherein M the second news centerings it is each
A second news centering includes belonging to two news of same event, N-M the second news to each of the second news
Centering includes belonging to two news of different event, and N, M are the positive integer more than or equal to 1, and N is greater than M.
Wherein, step 201 specifically includes the following steps:
Step 1: clustering to multiple news to be processed, event base is obtained.
Step 2: receiving markup information, wherein markup information includes to the news in each of event base event cluster
Update information whether current event cluster is belonged to the news in each of event base event cluster and according to markup information
It is modified, obtains revised event base.
Step 3: obtaining M the second news pair from revised event base.
Step 4: inquiring multiple news relevant to keyword from preset database according to preset keyword.
Step 5: will multiple news relevant to keyword, composition N-M the second news pair.
Wherein, the 5th step specifically includes: according to the descending of the matching degree with keyword, to relevant to keyword multiple
News is ranked up processing, multiple news after being sorted；Belong to different event in multiple news after choosing sequence
News, the multiple news selected；The multiple news that will be selected, N-M the second news pair of composition.
In the present embodiment, specifically, the executing subject of the embodiment of the present application can be terminal or server or
The processing unit or equipment of news information or other can execute the device or equipment of the embodiment of the present application, the application is not done
Limitation.
Need to obtain N number of second news pair to be trained, wherein M the second news to each of the second news pair
In include belonging to two news of same event, N-M the second news to each of the second news centering include ownership
In two news of different event.
Specifically, for obtaining M the second news for, first step is to get multiple news.
Second step is to be clustered using clustering method to multiple news, obtain event base, wherein wrap in event base
Multiple event clusters are included, include at least one news in each event cluster, each event cluster is corresponding with a kind of event.
Third step is to receive the markup information that user sends；According to markup information, to the news in each event cluster
Whether belong to current event cluster to be modified, obtains revised event base.Specifically, artificial to each event cluster
News is marked two-by-two, obtains multiple second news pair, and then marks out and whether the news of news centering belongs to currently
Event cluster, wherein the news of the second news centering can intersect.So as to obtain multiple second news pair, each
The news of second news centering must belong to the same event；Multiple second news pair are stored into event base.
Four steps is, to obtain out M the second news pair in revised event base.To M the second news
To each of the second news centering include two news, the news of each the second news centering is to belong to similar events
's.
For example, M=3,3 the second news pair, respectively the second news are to a, the second news to b, the second news pair
c；Second news belongs to event a to two news that a includes, and the second news is to belong to event b to b two news for including
, the second news belongs to event c to two news that c includes.
For obtaining N-M the second news for, first step is to obtain preset keyword, wherein in keyword
It include name entity and verb.
Second step is, according to keyword, is inquired, is obtained relevant to keyword in the database mature from one
Multiple news.Wherein it is possible to be matched according to keyword with the title of news, and then obtain relevant to keyword multiple new
It hears.
Third step is, according to the descending of the matching degree with keyword, to carry out to multiple news relevant to keyword
Sequence processing, multiple news after being sorted, wherein the sequence of higher news is more forward with the matching degree of keyword.
Four steps is, chooses the news of R before sorting, and event corresponding to this R news be it is different, into
And the R news not comprising similar events before selecting.Wherein, R is positive integer.
5th step is not include obtained R the news of similar events, carries out cross match two-by-two, obtains N-M
Second news pair.
For example, 3 news are selected, respectively news 1, news 2 and news 3, news 1 belongs to event a, and news 2 belongs to
Belong to event c in event b news 3；This 3 news are subjected to cross match two-by-two, obtain 3 the second news pair, respectively the
Two news are to d, the second news to e, the second news to f；Second news includes news 1 and news 2 to d, and news 1 belongs to event a,
News 2 belongs to event b；Second news includes news 1 and news 3 to e, and news 1 belongs to event a, and news 3 belongs to event c；Second
News includes news 2 and news 3 to f, and news 2 belongs to event b, and news 3 belongs to event c.
Step 202, to M the second news to and N-M the second news to feature calculation is carried out, obtain at least one feature
Information.
In the present embodiment, specifically, to N number of second news got to each of the second news pair, carry out
Feature calculation obtains one or more characteristic informations.For example, characteristic information: term vector side can be calculated in the following ways
Formula, TF-IDF (Term Frequency-Inverse Document Frequency, abbreviation TF-IDF) technology, name entity
Identification technology, verb identification technology, dependency analysis technology.
Wherein, one or more characteristic informations include the frequency of occurrence of keyword, the classification of news, the length of news,
Etc..For example, the title to news segments, the frequency of occurrence of keyword is obtained, the frequency of occurrence of keyword is as a kind of
Characteristic information.
M the second news are input to wait instruct, N-M the second news pair and at least one characteristic information by step 203
In two experienced disaggregated models, preset two disaggregated model is obtained.
In the present embodiment, specifically, by N number of second news to and each the second news pair characteristic information, input
Two disaggregated models into two disaggregated models to be trained, and then after being trained.
Step 204 obtains news agregator to be processed, wherein includes at least one news in news agregator.
In the present embodiment, specifically, obtaining news agregator to be processed first, wherein in news agregator to be processed
Including at least one news to be processed.
News agregator is input in two disaggregated models by step 205, to filter out the news of non-event, after obtaining processing
News agregator, wherein in news agregator that treated include news to be processed.
In the present embodiment, specifically, news agregator to be processed is input in two disaggregated models after training, in turn
The news for filtering out non-event, the news agregator that obtains that treated, in news agregator that treated news be all belong to it is a certain
A event.
Then, for each news to be processed, the step 206-211 of the embodiment of the present application is executed.
Step 206, from preset event base, determine similar to news to be processed at least one similar news,
In, it include at least one event cluster in event base, each event cluster is corresponding with a kind of event, has extremely in each event cluster
Few news, the similar news of each of at least one similar news are belonging respectively to different events.
In the present embodiment, it specifically, this step may refer to the step 102 of Fig. 1, repeats no more.
Step 207, by news to be processed, news similar with each is combined respectively, constitutes the first different news pair,
Wherein, the first news centering includes news and a similar news to be processed.
In the present embodiment, it specifically, this step may refer to the step 103 of Fig. 1, repeats no more.
Step 208, by each first news to being input in preset two disaggregated model, export each first news to category
In the probability value of similar events.
Wherein, step 208 specifically includes: by each first news to being input in preset two disaggregated model, exporting each
A first news is to the probability value for belonging to the event that each similar news of news centering is belonged to.
In the present embodiment, specifically, being exported by each first news in two disaggregated models after being input to training
The probability value that each first news belongs to the similar news for belonging to the news centering.Specifically, by each first
News in two disaggregated models after being input to training, exporting each news to whether belonging to the probability value of each event cluster,
And then obtain each news to whether belong to the probability value of each event to get to each news to belong to a certain kind it is identical
First probability value of event cluster, each news are to the second probability value for being not belonging to a certain similar events cluster.
For example, for news A to be processed, similar news 1 under event cluster a, similar new under event cluster b is obtained
Hear 2, the similar news 3 under event cluster c, the similar news 4 under event cluster d, the similar news 5 under event cluster e；Event cluster a's
Event is event a, and the event of event cluster b is event b, and the event of event cluster c is event c, and the event of event cluster d is event d, thing
The event of part cluster e is event e；News A and similar news 1 are formed into the first news to a, by news A and the composition of similar news 2 the
One news forms the first news to c to b, by news A and similar news 3, and news A and similar news 4 are formed the first news pair
News A and similar news 5 are formed the first news to e by d.By above-mentioned each first news to being input in two disaggregated models, obtain
Belong to the probability value a of the event a of similar news 1 to a to the first news, the first news belongs to the event b of similar news 2 to b
Probability value b, the first news belong to the probability value c of the event c of similar news 3 to c, and the first news belongs to the thing of similar news 4 to d
The probability value d of part d, the first news belong to the probability value e of the event e of similar news 5 to e.Also, the first news can also be obtained
The probability value of event b, event c, event d, event e are belonging respectively to a, the first news is belonging respectively to event a, event c, thing to b
The probability value of part d, event e, the first news are belonging respectively to the probability value of event a, event b, event d, event e, the first news to c
The probability value of event a, event b, event c, event e are belonging respectively to d, the first news is belonging respectively to event a, event b, thing to e
The probability value of part c, event d.
Step 209, if it is determined that maximum probability value be greater than preset threshold, it is determined that the maximum news centering of probability value is similar
The event that news is belonged to is event corresponding to news to be processed.
In the present embodiment, specifically, after step 208, however, it is determined that maximum probability value is greater than preset threshold, then will
The event that the similar news of the maximum news centering of probability value is belonged to, as event corresponding to news to be processed.
For example, according to the citing of step 108, probability value b is maximum, and probability value b is greater than preset threshold, it is determined that
The event of the news A of processing is event b.
Step 210, if it is determined that maximum probability value is less than or equal to preset threshold, then for news to be processed create one it is new
Event, and by new events as event corresponding to news to be processed.
In this step, specifically, after step 208, however, it is determined that maximum probability value is both less than equal to preset threshold,
Then it is found that the probability value of each the first news pair is very low, then can be to create a new events for news to be processed.
For example, according to the citing of step 108, probability value a, probability value b, probability value c, probability value d and probability value e are
It is very low, then an event cluster f is created for news A to be processed, determines that the event of news A to be processed is event f.
Step 211, by news to be processed, be put into corresponding to event corresponding to news to be processed in event base
In event cluster.
In this step, specifically, after step 209 or step 210, the event for determining news to be processed it
Afterwards, can determine with event cluster corresponding to the event of news to be processed, news to be processed is then put into event
In the event cluster in library.And then expand event base.
For example, according to the citing of step 209, however, it is determined that the event of news A to be processed is event b, event b and thing
Part cluster b is corresponding, then news A to be processed is put into the event cluster b of event base.
Again for example, according to the citing of step 210, create an event cluster f if news A to be processed, determine to
The event of the news A of processing is event f.An event cluster f then is increased newly for event base, is then put into news A to be processed
In the event cluster f of event base.
The embodiment of the present application, by obtaining news to be processed；From preset event base, determining and to be processed news
At least one similar similar news, wherein include at least one event cluster, each event cluster and a kind of event in event base
It is corresponding, there is at least one news, the similar news of each of at least one similar news belongs to respectively in each event cluster
In different events；By news to be processed, news similar with each is combined respectively, constitutes the first different news pair,
In, the first news centering includes news and a similar news to be processed；By each first news to being input to preset two points
In class model, event corresponding to news to be processed is obtained, wherein two disaggregated models are N number of the using the event that is labelled with
Two news are to being trained.Since two disaggregated models are using being labelled with the second news of event to being trained
It arrives, so that two disaggregated models are a kind of models for having supervision；Using there is the model of supervision to the first news to identifying, into
And accurately determine event corresponding to the news to be processed of the first news centering；What can be directed to divides news
Class improves the verification and measurement ratio for belonging to the news of the same event.And it is possible to filter out the news of the non-event in news, i.e.,
Filter out a large amount of noise information；Guarantee that the news belonged in the same event cluster is to belong to the event of same type, mentions
The high accuracy rate and recall rate of event monitoring；Then, only the news comprising event is presented to the user, needs to know in user
When certain a kind of event, the news screening that user carries out the long period is not needed, user experience is improved.
Fig. 3 is a kind of structural schematic diagram of the processing unit of news information provided by the embodiments of the present application, as shown in figure 3,
Device provided in this embodiment, comprising:
First acquisition unit 31, for obtaining news to be processed.
First determination unit 32, for from preset event base, determine it is similar with news to be processed at least one
Similar news, wherein include at least one event cluster in event base, each event cluster is corresponding with a kind of event, each thing
There is at least one news, the similar news of each of at least one similar news is belonging respectively to different events in part cluster.
First processing units 33 are constituted different for by news to be processed, news similar with each to be combined respectively
First news pair, wherein the first news centering includes news and a similar news to be processed.
Second determination unit 34, for being input in preset two disaggregated model, obtaining to be processed each first news
News corresponding to event, wherein two disaggregated models are using being labelled with N number of second news of event to being trained to obtain
, M the second news to each of the second news centering include belonging to two news of same event, N-M second
News to each of the second news centering include belonging to two news of different event, N, M be it is just whole more than or equal to 1
Number, N are greater than M.
The processing unit of news information provided in this embodiment is same as realizing the news information that aforementioned any embodiment provides
Processing method in technical solution, realization principle is similar, repeats no more.
The embodiment of the present application, by obtaining news to be processed；From preset event base, determining and to be processed news
At least one similar similar news, wherein include at least one event cluster, each event cluster and a kind of event in event base
It is corresponding, there is at least one news, the similar news of each of at least one similar news belongs to respectively in each event cluster
In different events；By news to be processed, news similar with each is combined respectively, constitutes the first different news pair,
In, the first news centering includes news and a similar news to be processed；By each first news to being input to preset two points
In class model, event corresponding to news to be processed is obtained, wherein two disaggregated models are N number of the using the event that is labelled with
Two news are to being trained.Since two disaggregated models are using being labelled with the second news of event to being trained
It arrives, so that two disaggregated models are a kind of models for having supervision；Using there is the model of supervision to the first news to identifying, into
And accurately determine event corresponding to the news to be processed of the first news centering；What can be directed to divides news
Class improves the verification and measurement ratio for belonging to the news of the same event.
Fig. 4 is the structural schematic diagram of the processing unit of another news information provided by the embodiments of the present application, shown in Fig. 3
On the basis of embodiment, as shown in figure 4, in device provided in this embodiment, the second determination unit 34, comprising:
Output module 341, for each first news to being input in preset two disaggregated model, is exported each first
News is to the probability value for belonging to similar events.
First determining module 342 is used for if it is determined that maximum probability value is greater than preset threshold, it is determined that probability value is maximum
The event that the similar news of news centering is belonged to is event corresponding to news to be processed.
Second determination unit 34, further includes:
Second determining module 343 is used for if it is determined that maximum probability value is less than or equal to preset threshold, is then to be processed new
One new events of creation are heard, and by new events as event corresponding to news to be processed.
Output module 341, is specifically used for: by each first news to being input in preset two disaggregated model, exporting each
A first news is to the probability value for belonging to the event that each similar news of news centering is belonged to.
First acquisition unit 31, comprising:
First obtains module 311, for obtaining news agregator to be processed, wherein includes at least one in news agregator
News.
Filtering module 312, to filter out the news of non-event, is obtained for news agregator to be input in two disaggregated models
To treated news agregator, wherein include news to be processed in news agregator that treated.
Device provided by the embodiments of the present application, further includes:
Second acquisition unit 41, for obtaining M the second news pair.
Third acquiring unit 42, for N-M the second news pair.
Computing unit 43, for M the second news to and N-M the second news to feature calculation is carried out, obtain at least
A kind of characteristic information.
Output unit 44, for M the second news are defeated to, N-M the second news pair and at least one characteristic information
Enter into two disaggregated models to be trained, obtains preset two disaggregated model.
Second acquisition unit 41, comprising:
Cluster module 411 obtains event base for clustering to multiple news to be processed.
Receiving module 412, for receiving markup information, wherein markup information includes to each of event base event
The update information of news in cluster, and according to markup information, whether the news in each of event base event cluster is belonged to
Current event cluster is modified, and obtains revised event base.
Second obtains module 413, for obtaining M the second news pair from revised event base.
Third acquiring unit 42, comprising:
Enquiry module 421, for being inquired from preset database relevant to keyword according to preset keyword
Multiple news.
Comprising modules 422, for will multiple news relevant to keyword, composition N-M the second news pair.
Comprising modules 422, are specifically used for: according to the descending of the matching degree with keyword, to relevant to keyword more
A news is ranked up processing, multiple news after being sorted；Belong to different event in multiple news after choosing sequence
News, the multiple news selected；The multiple news that will be selected, N-M the second news pair of composition.
Device provided by the embodiments of the present application, further includes:
The second processing unit 45, for the second determination unit 34 by each first news to be input to it is preset two classification mould
In type, after obtaining event corresponding to news to be processed, news to be processed is put into be processed new in event base
It hears in event cluster corresponding to corresponding event.
The processing unit of news information provided in this embodiment is same as realizing the news information that aforementioned any embodiment provides
Processing method in technical solution, realization principle is similar, repeats no more.
The embodiment of the present application, by obtaining news to be processed；From preset event base, determining and to be processed news
At least one similar similar news, wherein include at least one event cluster, each event cluster and a kind of event in event base
It is corresponding, there is at least one news, the similar news of each of at least one similar news belongs to respectively in each event cluster
In different events；By news to be processed, news similar with each is combined respectively, constitutes the first different news pair,
In, the first news centering includes news and a similar news to be processed；By each first news to being input to preset two points
In class model, event corresponding to news to be processed is obtained, wherein two disaggregated models are N number of the using the event that is labelled with
Two news are to being trained.Since two disaggregated models are using being labelled with the second news of event to being trained
It arrives, so that two disaggregated models are a kind of models for having supervision；Using there is the model of supervision to the first news to identifying, into
And accurately determine event corresponding to the news to be processed of the first news centering；What can be directed to divides news
Class improves the verification and measurement ratio for belonging to the news of the same event.And it is possible to filter out the news of the non-event in news, i.e.,
Filter out a large amount of noise information；Guarantee that the news belonged in the same event cluster is to belong to the event of same type, mentions
The high accuracy rate and recall rate of event monitoring；Then, only the news comprising event is presented to the user, needs to know in user
When certain a kind of event, the news screening that user carries out the long period is not needed, user experience is improved.
Fig. 5 is a kind of structural schematic diagram for controlling equipment provided by the embodiments of the present application, as shown in figure 5, the control equipment,
It include: transmitter 71, receiver 72, memory 73 and processor 74；
Memory 73 is for storing computer instruction；Computer instruction of the processor 74 for run memory 73 to store is real
Existing previous embodiment provides the technical solution of the obstacle classification method based on unmanned vehicle of any implementation.
The application also provides a kind of storage medium, comprising: readable storage medium storing program for executing and computer instruction, computer instruction storage
In readable storage medium storing program for executing；The obstacle based on unmanned vehicle for any implementation that computer instruction provides for realizing previous example
The technical solution of object classification method.
Above-mentioned control equipment in the specific implementation, it should be understood that processor 74 can be central processing unit (English:
Central Processing Unit, referred to as: CPU), can also be other general processors, digital signal processor (English:
Digital Signal Processor, referred to as: DSP), specific integrated circuit (English: Application Specific
Integrated Circuit, referred to as: ASIC) etc..General processor can be microprocessor or the processor is also possible to
Any conventional processor etc..The step of method in conjunction with disclosed in the embodiment of the present application, can be embodied directly in hardware processor
Execute completion, or in processor hardware and software module combination execute completion.
Those of ordinary skill in the art will appreciate that: realize that all or part of the steps of above-mentioned each method embodiment can lead to
The relevant hardware of program instruction is crossed to complete.Program above-mentioned can be stored in a computer readable storage medium.The journey
When being executed, execution includes the steps that above-mentioned each method embodiment to sequence；And storage medium above-mentioned includes: read-only memory (English
Text: read-only memory, abbreviation: ROM), RAM, flash memory, hard disk, solid state hard disk, tape (English: magnetic
Tape), floppy disk (English: floppy disk), CD (English: optical disc) and any combination thereof.
Finally, it should be noted that the above various embodiments is only to illustrate the technical solution of the application, rather than its limitations；To the greatest extent
Pipe is described in detail the application referring to foregoing embodiments, those skilled in the art should understand that: its according to
So be possible to modify the technical solutions described in the foregoing embodiments, or to some or all of the technical features into
Row equivalent replacement；And these are modified or replaceed, each embodiment technology of the application that it does not separate the essence of the corresponding technical solution
The range of scheme.
Claims (22)
1. a kind of processing method of news information characterized by comprising
Obtain news to be processed；
From preset event base, at least one similar news similar to the news to be processed is determined, wherein the thing
It include at least one event cluster in part library, each described event cluster is corresponding with a kind of event, has in each described event cluster
There is at least one news, the similar news of each of at least one described similar news is belonging respectively to different events；
It combines the news to be processed with similar news described in each respectively, constitutes the first different news pair, wherein
The first news centering includes the news to be processed and the similar news；
By each first news to being input in preset two disaggregated model, thing corresponding to the news to be processed is obtained
Part, wherein two disaggregated model is N number of second news using the event that is labelled with to being trained, and M second new
Hear to each of the second news centering include belonging to two news of same event, N-M the second news centering it is every
One the second news centering includes belonging to two news of different event, and N, M are the positive integer more than or equal to 1, and N is greater than M.
2. the method according to claim 1, wherein each first news is classified to being input to preset two
In model, event corresponding to the news to be processed is obtained, comprising:
By each first news to being input in preset two disaggregated model, each described first news is exported to belonging to phase
With the probability value of event；
If it is determined that maximum probability value is greater than preset threshold, it is determined that the similar news of the maximum news centering of probability value is belonged to
Event is event corresponding to the news to be processed.
3. according to the method described in claim 2, it is characterized in that, the method, further includes:
If it is determined that maximum probability value is less than or equal to preset threshold, then a new events are created for the news to be processed, and
By the new events as event corresponding to the news to be processed.
4. according to the method described in claim 2, it is characterized in that, each first news is classified to being input to preset two
In model, each described first news is exported to the probability value for belonging to similar events, comprising:
By each first news to being input in preset two disaggregated model, it is every to belonging to export each described first news
The probability value for the event that the similar news of one news centering is belonged to.
5. the method according to claim 1, wherein described obtain news to be processed, comprising:
Obtain news agregator to be processed, wherein include at least one news in the news agregator；
The news agregator is input in two disaggregated model, to filter out the news of non-event, it is new to obtain that treated
Hear set, wherein include the news to be processed in treated the news agregator.
6. method according to claim 1-5, which is characterized in that the method, further includes:
Obtain the M the second news to the N-M the second news pair；
, to feature calculation is carried out, at least one feature letter is obtained to the N-M the second news to the M the second news
Breath；
The M the second news are input to wait instruct, the N-M the second news pair and at least one characteristic information
In two experienced disaggregated models, preset two disaggregated model is obtained.
7. according to the method described in claim 6, it is characterized in that, obtaining the M the second news pair, comprising:
Multiple news to be processed are clustered, the event base is obtained；
Receive markup information, wherein the markup information includes to the news in each of described event base event cluster
Update information, and according to the markup information, whether the news in each of described event base event cluster is belonged to currently
Event cluster is modified, and obtains revised event base；
The M the second news pair are obtained from the revised event base.
8. according to the method described in claim 6, it is characterized in that, obtaining the N-M the second news pair, comprising:
According to preset keyword, multiple news relevant to the keyword are inquired from preset database；
Will multiple news relevant to the keyword, form the N-M the second news pair.
9. according to the method described in claim 8, it is characterized in that, will multiple news relevant to the keyword, form institute
State N-M the second news pair, comprising:
According to the descending of the matching degree with keyword, processing is ranked up to multiple news relevant to the keyword, is obtained
Multiple news after to sequence；
The news for belonging to different event in multiple news after choosing the sequence, the multiple news selected；
By the multiple news selected, the N-M the second news pair are formed.
10. method according to claim 1-5, which is characterized in that by each first news to being input to
In preset two disaggregated model, after obtaining event corresponding to the news to be processed, further includes:
The news to be processed is put into corresponding to event corresponding to news to be processed described in the event base
In event cluster.
11. a kind of processing unit of news information characterized by comprising
First acquisition unit, for obtaining news to be processed；
First determination unit, for determining at least one phase similar with the news to be processed from preset event base
Like news, wherein include at least one event cluster in the event base, each described event cluster is corresponding with a kind of event, often
There is at least one news, the similar news of each of at least one described similar news belongs to respectively in one event cluster
In different events；
First processing units are constituted different for combining the news to be processed with similar news described in each respectively
The first news pair, wherein the first news centering includes the news to be processed and the similar news；
Second determination unit, it is described wait locate for being input in preset two disaggregated model, obtaining each first news
Event corresponding to the news of reason, wherein two disaggregated model is using being labelled with N number of second news of event to instructing
Get, M the second news to each of the second news centering include belonging to two news of same event, N-M
A second news to each of the second news centering include belonging to two news of different event, N, M be more than or equal to 1
Positive integer, N be greater than M.
12. device according to claim 11, which is characterized in that second determination unit, comprising:
Output module, for each first news to being input in preset two disaggregated model, is exported each described
One news is to the probability value for belonging to similar events；
First determining module is used for if it is determined that maximum probability value is greater than preset threshold, it is determined that the maximum news pair of probability value
In the event that is belonged to of similar news, be event corresponding to the news to be processed.
13. device according to claim 12, which is characterized in that second determination unit, further includes:
Second determining module, for if it is determined that maximum probability value is then the news to be processed less than or equal to preset threshold
A new events are created, and by the new events as event corresponding to the news to be processed.
14. device according to claim 12, which is characterized in that the output module is specifically used for:
By each first news to being input in preset two disaggregated model, it is every to belonging to export each described first news
The probability value for the event that the similar news of one news centering is belonged to.
15. device according to claim 11, which is characterized in that the first acquisition unit, comprising:
First obtains module, for obtaining news agregator to be processed, wherein includes that at least one is new in the news agregator
It hears；
Filtering module, to filter out the news of non-event, is obtained for the news agregator to be input in two disaggregated model
To treated news agregator, wherein include the news to be processed in treated the news agregator.
16. the described in any item devices of 1-15 according to claim 1, which is characterized in that described device, further includes:
Second acquisition unit, for obtaining the M the second news pair；
Third acquiring unit, for the N-M the second news pair；
Computing unit, for the M the second news to, to feature calculation is carried out, obtained with the N-M the second news to
A kind of few characteristic information；
Output unit, for by the M the second news to, the N-M the second news pair and at least one feature
Information input obtains preset two disaggregated model into two disaggregated models to be trained.
17. device according to claim 16, which is characterized in that the second acquisition unit, comprising:
Cluster module obtains the event base for clustering to multiple news to be processed；
Receiving module, for receiving markup information, wherein the markup information includes to each of event base event
The update information of news in cluster, and according to the markup information, to the news in each of described event base event cluster
Whether belong to current event cluster to be modified, obtains revised event base；
Second obtains module, for obtaining the M the second news pair from the revised event base.
18. device according to claim 16, which is characterized in that the third acquiring unit, comprising:
Enquiry module, for being inquired from preset database relevant to the keyword more according to preset keyword
A news；
Comprising modules, for will multiple news relevant to the keyword, form the N-M the second news pair.
19. device according to claim 18, which is characterized in that the comprising modules are specifically used for:
According to the descending of the matching degree with keyword, processing is ranked up to multiple news relevant to the keyword, is obtained
Multiple news after to sequence；
The news for belonging to different event in multiple news after choosing the sequence, the multiple news selected；
By the multiple news selected, the N-M the second news pair are formed.
20. the described in any item devices of 1-15 according to claim 1, which is characterized in that described device, further includes:
The second processing unit, for second determination unit by each first news to be input to it is preset two classification mould
In type, after obtaining event corresponding to the news to be processed, by the news to be processed, it is put into the event base
Described in event cluster corresponding to event corresponding to news to be processed.
21. a kind of control equipment characterized by comprising transmitter, receiver, memory and processor；
The memory is for storing computer instruction；The processor is used to run the computer of the memory storage
The described in any item methods of claims 1 to 10 are realized in instruction.
22. a kind of storage medium characterized by comprising readable storage medium storing program for executing and computer instruction, the computer instruction are deposited
Storage is in the readable storage medium storing program for executing；The computer instruction is for realizing the described in any item methods of claims 1 to 10.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811581267.9A CN109857859B (en) | 2018-12-24 | 2018-12-24 | News information processing method, device, equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811581267.9A CN109857859B (en) | 2018-12-24 | 2018-12-24 | News information processing method, device, equipment and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109857859A true CN109857859A (en) | 2019-06-07 |
| CN109857859B CN109857859B (en) | 2021-03-16 |
Family
ID=66892064
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811581267.9A Active CN109857859B (en) | 2018-12-24 | 2018-12-24 | News information processing method, device, equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109857859B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111460289A (en) * | 2020-03-27 | 2020-07-28 | 北京百度网讯科技有限公司 | News information pushing method and device |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20090070346A1 (en) * | 2007-09-06 | 2009-03-12 | Antonio Savona | Systems and methods for clustering information |
| CN103336847A (en) * | 2013-07-22 | 2013-10-02 | 厦门市美亚柏科信息股份有限公司 | Generation method and system for hot news tag |
| CN103870474A (en) * | 2012-12-11 | 2014-06-18 | 北京百度网讯科技有限公司 | News topic organizing method and device |
| CN105677894A (en) * | 2016-02-02 | 2016-06-15 | 清华大学 | Network event model based news event monitoring method and device |
| US20160275172A1 (en) * | 2015-03-18 | 2016-09-22 | Fujitsu Limited | Non-transitory computer-readable recording medium, data classifying method, and data classifying device |
| CN108734216A (en) * | 2018-05-22 | 2018-11-02 | 广东工业大学 | Classification of power customers method, apparatus and storage medium based on load curve form |
-
2018
- 2018-12-24 CN CN201811581267.9A patent/CN109857859B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20090070346A1 (en) * | 2007-09-06 | 2009-03-12 | Antonio Savona | Systems and methods for clustering information |
| CN103870474A (en) * | 2012-12-11 | 2014-06-18 | 北京百度网讯科技有限公司 | News topic organizing method and device |
| CN103336847A (en) * | 2013-07-22 | 2013-10-02 | 厦门市美亚柏科信息股份有限公司 | Generation method and system for hot news tag |
| US20160275172A1 (en) * | 2015-03-18 | 2016-09-22 | Fujitsu Limited | Non-transitory computer-readable recording medium, data classifying method, and data classifying device |
| CN105677894A (en) * | 2016-02-02 | 2016-06-15 | 清华大学 | Network event model based news event monitoring method and device |
| CN108734216A (en) * | 2018-05-22 | 2018-11-02 | 广东工业大学 | Classification of power customers method, apparatus and storage medium based on load curve form |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111460289A (en) * | 2020-03-27 | 2020-07-28 | 北京百度网讯科技有限公司 | News information pushing method and device |
| CN111460289B (en) * | 2020-03-27 | 2024-03-29 | 北京百度网讯科技有限公司 | News information pushing method and device |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109857859B (en) | 2021-03-16 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US20210182611A1 (en) | Training data acquisition method and device, server and storage medium | |
| US20180101594A1 (en) | System and method for news events detection and visualization | |
| CA2777506C (en) | System and method for grouping multiple streams of data | |
| CN106021362A (en) | Query picture characteristic representation generation method and device, and picture search method and device | |
| CN110222233B (en) | Video recommendation method and device, server and storage medium | |
| CA2765111C (en) | Method and system for estimating age of a user based on mass data | |
| CN108021708B (en) | Content recommendation method and device and computer readable storage medium | |
| CN112148881B (en) | Method and device for outputting information | |
| US10467255B2 (en) | Methods and systems for analyzing reading logs and documents thereof | |
| KR20130062442A (en) | Method and system for recommendation using style of collaborative filtering | |
| CN108664515B (en) | A kind of searching method and device, electronic equipment | |
| EP3103272A1 (en) | Method, system and apparatus for configuring a chatbot | |
| CN102426577A (en) | Information processing apparatus, information processing system, information processing method, and program | |
| CN111460195A (en) | Picture processing method and device, storage medium and electronic equipment | |
| US8301584B2 (en) | System and method for adaptive pruning | |
| GB2600369A (en) | Active learning for data matching | |
| CN110209916B (en) | Method and device for recommending point of interest images | |
| CN109857859A (en) | Processing method, device, equipment and the storage medium of news information | |
| CN111177564B (en) | Product recommendation method and device | |
| CN113158022A (en) | Service recommendation method, device, server and storage medium | |
| CN103038766B (en) | Logical operation system | |
| Gias et al. | Samplehst: Efficient on-the-fly selection of distributed traces | |
| CN111026940A (en) | Network public opinion and risk information monitoring system and electronic equipment for power grid electromagnetic environment | |
| CN114238062B (en) | Board card burning device performance analysis method, device, equipment and readable storage medium | |
| CN109446408A (en) | Retrieve method, apparatus, equipment and the computer readable storage medium of set of metadata of similar data |
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 |