CN107562912A - Sina weibo event recommendation method - Google Patents
Sina weibo event recommendation method Download PDFInfo
- Publication number
- CN107562912A CN107562912A CN201710816042.6A CN201710816042A CN107562912A CN 107562912 A CN107562912 A CN 107562912A CN 201710816042 A CN201710816042 A CN 201710816042A CN 107562912 A CN107562912 A CN 107562912A
- Authority
- CN
- China
- Prior art keywords
- mrow
- event
- user
- model
- msub
- 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
Abstract
The present invention discloses a kind of Sina weibo event recommendation method, and for the problem of social short text proposed algorithm degree of accuracy is not high at present, user model and the similarity of event vector are calculated by improved cosine angle algorithm;If similarity pushes the event higher than the threshold value of setting to user;And user model is updated by the newly arrived time in nearest a period of time, event latest development state can be tracked;Behavior is thumbed up with reference to user to be again updated user model, recommendation results is more conformed to user's expection;The present processes can recommend Sina weibo event with higher accuracy rate, model reasonably can be drifted about, and can timely respond to the feedback to user's recommendation results.
Description
Technical field
The invention belongs to Data Mining, more particularly to a kind of social networks text recommended technology.
Background technology
For microblogging as a kind of new communications media, development is swift and violent, has that spread speed is fast, interactive strong, information updating side
Just the features such as, have begun to have an immense impact on to social life, turn into one of main social networks communication media in China.Due to
People can outwardly be released news by various forms such as web, webpages whenever and wherever possible, and realization is shared immediately, increasingly
More people likes sharing information, exchange opinion on microblogging and showed emotion.Compared with traditional media, for many grave news
Event, easy to operate, the low threshold of microblogging determine that microblogging more can take up the commanding elevation of information issue.This point is in burst thing
Showed in part it is more prominent because any microblog users in love scene can be issued whole event information by mobile phone
Gone on to microblogging.For example, in November, 2009,4.4 grades of earthquakes occur for Xi'an, and microblogging was only just reported after 1 minute to the event
Road, and issue is after 15 minutes for the first time for national official website.
But with the popularization of microblogging, also bring the problem of some are new.Matter of utmost importance is information explosion, and the data of magnanimity are believed
Breath is flooded with internet, has brought serious problem of information overload.People are often difficult in face of the information of this magnanimity
Oneself desired data is found, and wants quickly and accurately to find then more difficult for oneself most important data.
In the past, people obtain information to web2.0 typically by the search engine of specialty, but this is there is also some problems, most important to ask
One of topic is that search engine needs user actively to go to inquire about, and it can not pushed information, real-time be high on one's own initiative so that Yong Huyou
Possible miss critical information.Web2.0 appearance so that everybody can participate in the issue of information, propagation and mistake by network
Filter, so as to reach the purpose of information sharing.The information push mode in this directed message source passes through search before although having overturned
Engine pulls the mode of information, but also compensate for the awkward situation that search engine currently faces well.
Method of the commending system as an acquisition of information, it studies the hobby of user from user, can be in user
It is intended in the case of obscuring, guiding user has found his pent-up demand, is pushed to his information interested, this acquisition of information side
Formula is that solve the very potential method of problem of information overload.The main task of commending system is the interest for accurately holding user
Point, using efficient proposed algorithm, to the possible event interested of its push.
Sina weibo has following feature as domestic most popular microblogging instrument:Blog article number of words be limited in 140 words with
Interior, data magnanimity, short essay person's character, text are Deletional, real-time, abundant social information.Due to the characteristic of microblog data, blog article
Form is not fixed, and many blog articles may not include effective information, and very big trouble is brought to handling, therefore currently for
The research of the commending system of this short text is still challenging.In order to reach good recommendation effect, exploitation is efficient
Proposed algorithm be particularly important.Current commending system is mostly text commending system, for this short text number of microblogging
According to commending system research it is deep not enough, its result of study is not met by practical application needs.
The content of the invention
In order to solve the above-mentioned technical problem, present applicant proposes a kind of Sina weibo event recommendation method, in real time amendment to use
Family model, the recommendation degree of accuracy of microblogging event recommendation system is improved, improve Consumer's Experience.
The technical solution adopted by the present invention is:Sina weibo event recommendation method, including:
S1, the similarity between user model and event vector calculated using improved cosine angle algorithm, if similarity
More than threshold value, then by the event recommendation to user;Otherwise do not recommend;
S2, according to the recommendation event of arrival event database in nearest duration K user model is updated;
S3, user model is updated according to thumbing up event by user.
Further, improved cosine angle algorithm is specially:
Wherein, sameWordNum represents user model A and event model B identical keyword numbers;min(|A|,|B|)
Represent dimension minimum in user model A and event model B;waiRepresent weight corresponding to Feature Words ai in user model A;wbj
Represent weight corresponding to Feature Words bj in event model B.
Further, the user model is extracted from customer data base.
Further, the event vector extracts from event database.
Further, step S2 is specially:
S21, when there is new recommendation event to reach in event database, then extract the recommendation thing reached in nearest duration K
Part;
Weight is more than the Feature Words addition user model of first threshold in each recommendation event of S22, selecting step S21 extraction
In;
High frequency vocabulary in S23, selection active user's aspect of model word is as new user model.
Further, step S3 is specially:When there is new event to be thumbed up, then record is thumbed up the ID of event, according to ID
Corresponding event is searched from event database, extracts the high frequency vocabulary of the event.
Beneficial effects of the present invention:The Sina weibo event recommendation method of the application, passes through improved cosine angle algorithm
Calculate user model and the similarity of event vector;If similarity pushes the event higher than the threshold value of setting to user;And
User model is updated by the newly arrived time in nearest a period of time, event latest development shape can be tracked
State;Behavior is thumbed up with reference to user to be again updated user model, recommendation results is more conformed to user's expection;The application's
Method can recommend Sina weibo event with higher accuracy rate, and model reasonably can be drifted about, and can be timely
Respond the feedback to user's recommendation results.
Brief description of the drawings
Fig. 1 is the protocol procedures figure of the application;
Fig. 2 is model drift workflow;
Fig. 3 is user feedback more new technological process.
Embodiment
For ease of skilled artisan understands that the technology contents of the present invention, enter one to present invention below in conjunction with the accompanying drawings
Step explaination.
It is the protocol procedures figure of the application as shown in Figure 1, the technical scheme of the application is:Sina weibo event recommendation side
Method, including:
S1, the similarity between user model and event vector calculated using improved cosine angle algorithm, if similarity
More than threshold value, then by the event recommendation to user;Otherwise do not recommend;
S2, according to the recommendation event of arrival event database in nearest duration K user model is updated;
S3, user model is updated according to thumbing up event by user.
Step S1 is specially:Classical cosine angle algorithmic formula is as follows:
Wherein, A, B represent user model vector sum event vector respectively, can represent as follows:
A={ (a1, wa1),(a2,wa2),(a3,wa3),……,(am,wam),}
B={ (b1, wb1),(b2,wb2),(b3,wb3),……,(bn,wbn),}
wa1Represent the weight corresponding to Feature Words a1 in user model A;B vectors are similarly.Abbreviation obtains:
Wherein, waiWith wbjThe condition of multiplication is characterized word ai=bj.
But if two vectorial identical words are more, then cosine value is larger.In view of user model and event vector
Dimension may it is larger, merely with morphology it is identical come calculate similarity unavoidably can cause recommend precision it is not high the problem of.Cause this
One of the reason for phenomenon is that the high Feature Words of weight and may not have the ability of division event in some events vector, as " in
State ", " U.S. " etc., and the slightly lower word of some weights is possible to be only the emphasis of event, such as " airplane crash ", " Golden Ball Award ".Cause
This, the application introduces an attenuation coefficient to improve recommendation precision, and the cosine angle algorithm after improvement is as follows:
Wherein, sameWordNum represents user model A and event model B identical keyword numbers;min(|A|,|B|)
Represent dimension minimum in user model A and event model B;waiRepresent weight corresponding to Feature Words ai in user model A;wbj
Represent weight corresponding to Feature Words bj in event model B.
Introduce after attenuation coefficient, if only having between vector, a small amount of keyword is identical, and its similarity can significantly decay,
As long as setting suitable threshold value, the setting of threshold value is not fixed in the application;General suitable threshold value, which can be realized, is recommended
Result meet expection, then the setting for illustrating the threshold value is suitable;Otherwise threshold value is readjusted.Can largely improves
Recommend precision.In addition to attenuation coefficient is introduced, the application has also used other two kinds and has improved the method for recommending precision simultaneously.It is first, hard
Property regulation same keyword number be more than and how many just recommended.Generally, the keyword of user's input will not be too many, about
5 or so, then the application setting just carries out similarity meter when user model and event vector at least three same keywords
Calculate.When user's input word number has big change, the threshold value can make corresponding adjustment.Second, in order to avoid same word not
The negative effect brought with morphology, when similarity is calculated, the application is extracted the stem of each keyword to be calculated.
After obtaining recommendation event, customer data base is deposited into, daily record is recommended in generation., can in the case of less demanding
To represent an event with the summary of event, the summary is pushed to user.If user needs to read original blog article, need from
Extraction and the maximally related blog article of user model in event.
Most interested blog article is extracted, it is necessary to be pre-processed to blog article, is segmented, stem reduction, if the blog article bag
The maximum word of weight in list containing same keyword, then the blog article may be that user is most interested in.
Step S2 is specially:The main task of model drift is user model is repaiied automatically over time
Just, its object is to real-time tracking event focus, event trend is grasped.
User model represents the point of interest of user, and it is generally also the epitome of an event, and simply user is closed with some
Keyword summarises this event.Over time, event may have new development, and its focus vocabulary is also varied from.For
From this change of motion tracking, ensure that user can receive up-to-date information, the application devises model shift module.
It is to correct user model in place of model drift core, newest focus vocabulary is added to user model, and delete
Out-of-date keyword in model.Its workflow is as shown in Figure 2.
With recommending module above, the application extracts user model from customer data base, and newest thing is extracted from event database
Part.It is worth noting that, extraction is the event reached within a nearest hour, and it is under the user model to extract trigger point
There is new recommendation event to reach.This means, when active user's model has new recommendation event, nearest one is generally extracted under the model
(the duration K i.e. in the application is a hour to all events being recommended within hour, and K value can also be other values, only
It is drift band difference;However, to ensure that news is ageing, it is not recommended that takes too big value), generate shift vectors.So do
Purpose be Drift Process is tended towards stability, avoid the occurrence of the too fast situation of drifting about.If drift band is too big, may with it is first
Beginning model greatly differs from each other, and influences Consumer's Experience.On the extraction of characteristic vector, also there are some noticeable places.It is one small
When within all events, the shift vectors of generation may be very big, far beyond the dimension of initial user model, in order to exclude
The minimum Feature Words of weight and avoid initial user model from excessively being watered down, choose the height in active user's aspect of model word
Frequency vocabulary is as new user model.
The affair character vector of extraction is defined to 20 words by the present embodiment, its a part of word of weight highest is only taken, by this
A little words are added to user model, and intercept after renewal before weight 20 word as new user model.Particularly, in order to ensure to use
Family is originally inputted the influence power of keyword, and the user model after renewal is divided into two parts by the application --- it is originally inputted key
Word, the keyword newly added, weight 0.5 is accounted for per part.The weight that both can guarantee that and be originally inputted keyword is so done, and can adds
Add the newest focus word of event and delete out-of-date Feature Words.
Likewise, the user model after drift is stored in customer data base, generation drift daily record.
Step S3 is specially:The main purpose of user feedback renewal is timely reception field feedback, according to user's
Hobby, amendment is made to user model.User reflects whether user is satisfied with to current results for the feedback of recommendation results,
It is the most important reference information modified to recommendation.User feedback more new technological process is as shown in Figure 3.
User feedback, most direct mode exactly " thumb up ".When user is to an event or a blog article interested, he
The event or blog article can be thumbed up, system identification to the behavior that thumbs up simultaneously will be thumbed up event id deposit customer data base.Utilize
This thumbs up information, can obtain the newest point of interest of user in time, updates user model.
Thumbed up if any new event, then user model and events of interest ID are extracted from customer data base, according to this
ID searches corresponding event in event database, extracts the high frequency vocabulary of the event;After high frequency vocabulary extracts, by high frequency
The weight of word frequency highest word is arranged to weight maximum in original subscriber's model in vocabulary, and remaining high frequency words weight is adjusted in proportion
It is whole.Finally, then to the user model after whole renewal it is normalized.So-called high frequency vocabulary, herein, is defined as word frequency
More than the Feature Words of corresponding event blog article number, this word has certain representative meaning for the event, it is clear that the weight of these words
Higher, renewal can effectively reflect that the subjective of user is liked to can occupy larger proportion when user model.
Drifted about with model, preceding 20 keywords of the user model after interception renewal do normalized, still to be originally inputted pass
Keyword accounts for weight 0.5.Finally, the user model after renewal is stored in customer data base, generates Update log.
One of ordinary skill in the art will be appreciated that embodiment described here is to aid in reader and understands this hair
Bright principle, it should be understood that protection scope of the present invention is not limited to such especially statement and embodiment.For ability
For the technical staff in domain, the present invention can have various modifications and variations.Within the spirit and principles of the invention, made
Any modification, equivalent substitution and improvements etc., should be included within scope of the presently claimed invention.
Claims (6)
1. Sina weibo event recommendation method, it is characterised in that including:
S1, the similarity between user model and event vector calculated using improved cosine angle algorithm, if similarity is more than
Threshold value, then by the event recommendation to user;Otherwise do not recommend;
S2, according to the recommendation event of arrival event database in nearest duration K user model is updated;
S3, user model is updated according to thumbing up event by user.
2. Sina weibo event recommendation method according to claim 1, it is characterised in that improved cosine angle algorithm tool
Body is:
<mrow>
<mi>cos</mi>
<mrow>
<mo>(</mo>
<mi>A</mi>
<mo>,</mo>
<mi>B</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mfrac>
<mrow>
<mi>s</mi>
<mi>a</mi>
<mi>m</mi>
<mi>e</mi>
<mi>W</mi>
<mi>o</mi>
<mi>r</mi>
<mi>d</mi>
<mi>N</mi>
<mi>u</mi>
<mi>m</mi>
</mrow>
<mrow>
<mi>min</mi>
<mrow>
<mo>(</mo>
<mo>|</mo>
<mi>A</mi>
<mo>|</mo>
<mo>,</mo>
<mo>|</mo>
<mi>B</mi>
<mo>|</mo>
<mo>)</mo>
</mrow>
</mrow>
</mfrac>
<mo>&times;</mo>
<mfrac>
<mrow>
<msub>
<mo>&Sigma;</mo>
<mrow>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
<mo>&Element;</mo>
<mrow>
<mo>(</mo>
<mn>0</mn>
<mo>,</mo>
<mi>max</mi>
<mo>(</mo>
<mrow>
<mi>m</mi>
<mo>,</mo>
<mi>n</mi>
</mrow>
<mo>)</mo>
<mo>)</mo>
</mrow>
</mrow>
</msub>
<msub>
<mi>w</mi>
<mrow>
<mi>a</mi>
<mi>i</mi>
</mrow>
</msub>
<msub>
<mi>w</mi>
<mrow>
<mi>b</mi>
<mi>j</mi>
</mrow>
</msub>
</mrow>
<mrow>
<msqrt>
<mrow>
<msubsup>
<mo>&Sigma;</mo>
<mrow>
<mi>i</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>m</mi>
</msubsup>
<msup>
<msub>
<mi>w</mi>
<mrow>
<mi>a</mi>
<mi>i</mi>
</mrow>
</msub>
<mn>2</mn>
</msup>
</mrow>
</msqrt>
<mo>&times;</mo>
<msqrt>
<mrow>
<msubsup>
<mo>&Sigma;</mo>
<mrow>
<mi>j</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>m</mi>
</msubsup>
<msup>
<msub>
<mi>w</mi>
<mrow>
<mi>b</mi>
<mi>j</mi>
</mrow>
</msub>
<mn>2</mn>
</msup>
</mrow>
</msqrt>
</mrow>
</mfrac>
</mrow>
Wherein, sameWordNum represents user model A and event model B identical keyword numbers;Min (| A |, | B |) represent
Minimum dimension in user model A and event model B;waiRepresent weight corresponding to Feature Words ai in user model A;wbjRepresent
Weight corresponding to Feature Words bj in event model B.
3. Sina weibo event recommendation method according to claim 2, it is characterised in that the user model is from number of users
According to being extracted in storehouse.
4. Sina weibo event recommendation method according to claim 3, it is characterised in that the event vector is from event number
According to being extracted in storehouse.
5. Sina weibo event recommendation method according to claim 1, it is characterised in that step S2 is specially:
S21, when there is new recommendation event to reach in event database, then extract the recommendation event reached in nearest duration K;
Weight is more than in the Feature Words addition user model of first threshold in each recommendation event of S22, selecting step S21 extraction;
High frequency vocabulary in S23, selection active user's aspect of model word is as new user model.
6. Sina weibo event recommendation method according to claim 1, it is characterised in that step S3 is specially:It is new when having
Event thumbed up, then record is thumbed up the ID of event, is searched correspondence event from event database according to ID, is extracted the event
High frequency vocabulary.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710816042.6A CN107562912B (en) | 2017-09-12 | 2017-09-12 | Sina microblog event recommendation method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710816042.6A CN107562912B (en) | 2017-09-12 | 2017-09-12 | Sina microblog event recommendation method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN107562912A true CN107562912A (en) | 2018-01-09 |
CN107562912B CN107562912B (en) | 2021-08-27 |
Family
ID=60980471
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710816042.6A Active CN107562912B (en) | 2017-09-12 | 2017-09-12 | Sina microblog event recommendation method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107562912B (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110175293A (en) * | 2019-05-30 | 2019-08-27 | 北京小米智能科技有限公司 | A kind of method, apparatus and electronic equipment of determining news train of thought |
Citations (11)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101968802A (en) * | 2010-09-30 | 2011-02-09 | 百度在线网络技术(北京)有限公司 | Method and equipment for recommending content of Internet based on user browse behavior |
US20110145822A1 (en) * | 2009-12-10 | 2011-06-16 | The Go Daddy Group, Inc. | Generating and recommending task solutions |
CN102831234A (en) * | 2012-08-31 | 2012-12-19 | 北京邮电大学 | Personalized news recommendation device and method based on news content and theme feature |
CN103455485A (en) * | 2012-05-28 | 2013-12-18 | 中兴通讯股份有限公司 | Method and device for automatically updating user interest model |
CN103488705A (en) * | 2013-09-06 | 2014-01-01 | 电子科技大学 | User interest model incremental update method of personalized recommendation system |
CN103778260A (en) * | 2014-03-03 | 2014-05-07 | 哈尔滨工业大学 | Individualized microblog information recommending system and method |
CN104239512A (en) * | 2014-09-16 | 2014-12-24 | 电子科技大学 | Text recommendation method |
US20150120829A1 (en) * | 2013-10-30 | 2015-04-30 | At&T Intellectual Property I, L.P. | Context based communication management |
CN105989056A (en) * | 2015-02-06 | 2016-10-05 | 北京中搜网络技术股份有限公司 | Chinese news recommending system |
KR20170024257A (en) * | 2015-08-25 | 2017-03-07 | 건국대학교 산학협력단 | Method and apparatus for recommending personalized subject |
CN106777132A (en) * | 2016-12-18 | 2017-05-31 | 深圳市辣妈帮科技有限公司 | Data processing method and device |
-
2017
- 2017-09-12 CN CN201710816042.6A patent/CN107562912B/en active Active
Patent Citations (11)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20110145822A1 (en) * | 2009-12-10 | 2011-06-16 | The Go Daddy Group, Inc. | Generating and recommending task solutions |
CN101968802A (en) * | 2010-09-30 | 2011-02-09 | 百度在线网络技术(北京)有限公司 | Method and equipment for recommending content of Internet based on user browse behavior |
CN103455485A (en) * | 2012-05-28 | 2013-12-18 | 中兴通讯股份有限公司 | Method and device for automatically updating user interest model |
CN102831234A (en) * | 2012-08-31 | 2012-12-19 | 北京邮电大学 | Personalized news recommendation device and method based on news content and theme feature |
CN103488705A (en) * | 2013-09-06 | 2014-01-01 | 电子科技大学 | User interest model incremental update method of personalized recommendation system |
US20150120829A1 (en) * | 2013-10-30 | 2015-04-30 | At&T Intellectual Property I, L.P. | Context based communication management |
CN103778260A (en) * | 2014-03-03 | 2014-05-07 | 哈尔滨工业大学 | Individualized microblog information recommending system and method |
CN104239512A (en) * | 2014-09-16 | 2014-12-24 | 电子科技大学 | Text recommendation method |
CN105989056A (en) * | 2015-02-06 | 2016-10-05 | 北京中搜网络技术股份有限公司 | Chinese news recommending system |
KR20170024257A (en) * | 2015-08-25 | 2017-03-07 | 건국대학교 산학협력단 | Method and apparatus for recommending personalized subject |
CN106777132A (en) * | 2016-12-18 | 2017-05-31 | 深圳市辣妈帮科技有限公司 | Data processing method and device |
Non-Patent Citations (4)
Title |
---|
LEI-LEI SHI 等: "Event Detection and User Interest Discovering in Social Media Data Streams", 《IEEE ACCESS》 * |
余燕川: "基于个性化新闻推荐模型及算法的研究", 《中国优秀硕士学位论文全文数据库信息科技辑》 * |
刘东: "推特事件推荐方法研究", 《中国优秀硕士学位论文全文数据库信息科技辑》 * |
孙辉 等: "一种相似度改进的用户聚类协同过滤推荐算法", 《小型微型计算机系统》 * |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110175293A (en) * | 2019-05-30 | 2019-08-27 | 北京小米智能科技有限公司 | A kind of method, apparatus and electronic equipment of determining news train of thought |
CN110175293B (en) * | 2019-05-30 | 2021-01-29 | 北京小米智能科技有限公司 | Method and device for determining news venation and electronic equipment |
Also Published As
Publication number | Publication date |
---|---|
CN107562912B (en) | 2021-08-27 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN103955505B (en) | A kind of event method of real-time and system based on microblogging | |
US20180129749A1 (en) | Method, apparatus, and system for recommending real-time information | |
Xiao et al. | Deep-reinforcement-learning-based user profile perturbation for privacy-aware recommendation | |
CN104809117B (en) | Video data aggregation processing method, paradigmatic system and video search platform | |
Lau et al. | Microblog Retrieval Using Topical Features and Query Expansion. | |
CN103324665A (en) | Hot spot information extraction method and device based on micro-blog | |
CN106484829B (en) | A kind of foundation and microblogging diversity search method of microblogging order models | |
CN102955781A (en) | Method and device for figure search | |
CN103823893A (en) | User comment-based product search method and system | |
CN104102681A (en) | Microblog key event acquiring method and device | |
CN102622375A (en) | Intelligent matching system and method for third-party lawyer recommendations | |
CN103123649A (en) | Method and system for searching information based on micro blog platform | |
CN107145545A (en) | Top k zone users text data recommends method in a kind of location-based social networks | |
CN102637179B (en) | Method and device for determining lexical item weighting functions and searching based on functions | |
CN105868267A (en) | Modeling method for mobile social network user interests | |
CN103164424A (en) | Method and device for acquiring time-efficient words | |
CN102662986A (en) | System and method for microblog message retrieval | |
CN105100353A (en) | Method for performing address book grouping on newly-added contact of mobile terminal | |
CN109492082A (en) | Pull down words recommending method, device, electronic equipment and storage medium | |
CN106789572A (en) | A kind of instant communicating system and instant communication method for realizing self adaptation message screening | |
CN106951410B (en) | Generation method, device and the electronic equipment of dictionary | |
CN103020141A (en) | Method and equipment for providing searching results | |
CN109992729A (en) | A kind of tourism strategy recommended method | |
CN107562912A (en) | Sina weibo event recommendation method | |
CN101840438B (en) | Retrieval system oriented to meta keywords of source document |
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 |