CN107908616A - The method and apparatus of anticipation trend word - Google Patents
The method and apparatus of anticipation trend word Download PDFInfo
- Publication number
- CN107908616A CN107908616A CN201710969459.6A CN201710969459A CN107908616A CN 107908616 A CN107908616 A CN 107908616A CN 201710969459 A CN201710969459 A CN 201710969459A CN 107908616 A CN107908616 A CN 107908616A
- Authority
- CN
- China
- Prior art keywords
- word
- search
- burstiness
- term vector
- vector
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/279—Recognition of textual entities
- G06F40/284—Lexical analysis, e.g. tokenisation or collocates
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
- G06F16/9535—Search customisation based on user profiles and personalisation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/06—Buying, selling or leasing transactions
- G06Q30/0601—Electronic shopping [e-shopping]
- G06Q30/0631—Item recommendations
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Databases & Information Systems (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- Finance (AREA)
- General Engineering & Computer Science (AREA)
- Accounting & Taxation (AREA)
- Development Economics (AREA)
- Data Mining & Analysis (AREA)
- Economics (AREA)
- Marketing (AREA)
- Strategic Management (AREA)
- General Business, Economics & Management (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses a kind of method and apparatus of anticipation trend word, it is related to field of computer technology.One embodiment of this method includes：Determine the burstiness of the search term in predetermined amount of time；Burstiness is met that the search term of pre-defined rule is determined as trend word.The embodiment can quantify the trend degree of different search terms, and according to trend degree determination trend word, and independent of the value of time window, so as to more adding system, exactly anticipation trend word, valuable reference and guide data are provided for the business activity of different field.
Description
Technical field
The present invention relates to field of computer technology, more particularly to a kind of method and apparatus of anticipation trend word.
Background technology
In the case where user accesses website, search data or carries out the scene of shopping online etc. by electric business platform, search is closed
Keyword is the important entrance that user obtains data message.Regard the search behavior of a large number of users within a certain period of time as one
It is overall, change and the tendency of overall keyword search amount can be observed by certain method, so as to excavate search hot spot
With trend.That is, the search activities of a large number of users are put down for site information provider, search engine service provider or electric business
The network service platforms such as platform provide important statistical information：What everybody is being concerned about, highly desirable to browse or understand which letter
Breath.
The larger search term of volumes of searches may be considered current hot spot, and actually find that the early stage of search term becomes in time
Gesture is more meaningful, and market is more strong for the demand of the latter.Because when some words become hot spot, then take one
A little commercial activities or measure are late；Before some search terms become hot spot, find that the early stage of these words becomes in time
Gesture becomes more valuable and effect, predicts that upcoming hot spot can provide extremely valuable ginseng for each different commercial fields
Examine and instruct.If for example, writer Mo Yan obtain the Nobel Prize become hot spot before, just find search term " Mo Yan " become compared with
Obvious early stage trend word, which can effectively instruct commercial activity, for example instruct the Cai Xiao departments of electric business platform to adopt in time
Purchase the written books of writer Mo Yan；Either Guide Reading platform or reading application etc. push books of writer Mo Yan etc. in time.
In terms of the prediction of typical time sequence, some classical time series predicting models, such as autoregression model, autoregression
Moving average model(MA model), and difference ARMA model etc., may be used to predict the frequency change of time series data, but
It is for predicting keyword search amount trend, also there is huge challenge using these models.Although because these models
Keyword search amount can be predicted to a certain extent, but can not point out whether corresponding search term can become early stage trend.
The basic skills for being generally used for judging early stage trend is to utilize statistical information (such as：Average, standard deviation etc.) sentenced
Not, but suitable time window must be used as the case may be, flexibility is subject to larger limitation.The detection mode is recognized as
It is that relative change rate differentiates, i.e., the increase and decrease degree between time quantum is judged according to the relative change rate of keyword search value
Deng.
Usually, often there is climax, time of the climax in Query trend in some periods in Query (inquiry)
A pulse can be intuitively reflected as in sequence, specifically the surge of exactly some keyword search amounts and is die-offed, i.e.,
Relative change rate is higher.Explained from practical significance, the inquiry climax of Query very likely correspond to one in real world
Dependent event.Therefore, Query sequential trend-monitoring namely detects pulse in Query enquiry frequency time serieses.
For given l predicted value { q1, q2 ... ..., ql }, a Query time series is in time interval [b, e]
There are a trend event, and if only if：
(1)1≤b≤e≤l；
(2) value in time interval [b, e] forms a pulse in corresponding time series enough in statistical significance,
It is exactly the average value that these values are much larger than in time series.
The decision rule of sequential trend is that the moving average in region goes out greatly δ than the average value in whole time series
Standard deviation.During this rule of practical application, usually value is 3 to δ.The following describe specific sequential trend word detection method and main
Step.
INPUT Query enquiry frequency time serieses Q=q1, q2 ..., ql }
If OUTPUT exports corresponding time ordered interval there are trend.
Step 1：The moving average MAw of time series Q is calculated with sliding window size w；
Step 2：The statistic in whole time series is calculated, setting decision threshold is：
Threshold=mean (MAw)+δ * std (MAw)；
Step 3：The sequential for obtaining trend event is { ti|MAw(i)>threshold}。
In process of the present invention is realized, inventor has found that at least there are the following problems in the prior art：
1st, when the method for existing detection sequential trend word must use suitable according to the concrete condition of different search terms
Between window, the obtained result difference of different time windows is larger, causes detection accuracy relatively low, and flexibility be subject to it is larger
Limitation；
2nd, further, since at present when carrying out the detection of trend word, the high search term of similarity is cannot be distinguished by, such as " ice
The similarity of the word such as river in Henan Province leaching ", " ice cream ", " ice lolly ", " ice cream " is high, and and for example the word such as " man ", " for men ", " man " also has
High similarity.The prior art can only detect each word respectively, be accustomed to different, each user couple yet with user
The statement of same event also differs, so each word is detected respectively can cause search term Trend judgement inaccurate, Wu Fagen
The Hot Contents that user pays close attention to accurately are judged according to search term trend.
The content of the invention
In view of this, the embodiment of the present invention provides a kind of method and apparatus of anticipation trend word, can quantify different search
The trend degree of word, and according to trend degree determination trend word, and independent of the value of time window, so as to more be
System, exactly anticipation trend word, valuable reference and guide data are provided for the business activity of different field.
To achieve the above object, a kind of one side according to embodiments of the present invention, there is provided method of anticipation trend word.
A kind of method of anticipation trend word, including：Determine the burstiness of the search term in predetermined amount of time；By the burst
Degree meets that the described search word of pre-defined rule is determined as trend word.
Optionally it is determined that before the step of burstiness of search term in predetermined amount of time, further include：By semantic similarity
Meet that the word of predetermined threshold gathers for one kind, and determine the classification logotype per class word；Search term in predetermined amount of time is replaced
It is changed to the classification logotype.
Alternatively, semantic similarity being met, the word of predetermined threshold gathers includes for a kind of step：Obtain corpus data；
Participle operation is carried out to the corpus data；Generate the term vector of each word obtained after the participle operation；According to institute's predicate
Semantic similarity is met that the word of predetermined threshold gathers for one kind by vector.
Alternatively, semantic similarity is met that the word of predetermined threshold gathers for a kind of step bag according to the term vector
Include：Following operation is performed to each term vector successively, until all corresponding words of term vector are referred to corresponding classification,
The operation includes：Obtain term vector A；The similarity of the term vector A and the cluster centre vector per class word are calculated, by institute
The maximum stated in similarity is denoted as d, and g, the cluster centre are denoted as with the word classification of the similarity maximum of the term vector A
Vector is the mean vector per all term vectors in class word；If d meets predetermined threshold, by the corresponding words of the term vector A
It is referred in classification g；Otherwise, a classification h is created, and the corresponding words of the term vector A are referred in the classification h.
Alternatively, the calculation formula of the burstiness is：
Wherein,It is burstiness, t is current time,It is the speed at current time, Δ T1、ΔT2
It is the size for the time window chosen.
Alternatively, the calculation formula of the speed at the current time is：
Wherein,It is the speed at current time, XiIt is the number that current search word occurs in i-th of search instruction,
tiIt is the timestamp of i-th of search instruction, Δ T is the size for the time window chosen.
A kind of other side according to embodiments of the present invention, there is provided device of anticipation trend word.
A kind of device of anticipation trend word, including：Computing module, for determining the burst of the search term in predetermined amount of time
Degree；Determining module, the described search word for the burstiness to be met to pre-defined rule are determined as trend word.
Alternatively, cluster module is further included, is used for：, will before the burstiness for determining the search term in predetermined amount of time
Semantic similarity meets that the word of predetermined threshold gathers for one kind, and determines the classification logotype per class word；By in predetermined amount of time
Search term replace with the classification logotype.
Alternatively, the cluster module is additionally operable to：Obtain corpus data；Participle operation is carried out to the corpus data；It is raw
The term vector of each word obtained after into the participle operation；Semantic similarity is met by predetermined threshold according to the term vector
Word gathers for one kind.
Alternatively, the cluster module is additionally operable to：Following operation is performed to each term vector successively, until will be all
The corresponding word of term vector is referred to corresponding classification, and the operation includes：Obtain term vector A；Calculate the term vector A and every class
The similarity of the cluster centre vector of word, d is denoted as by the maximum in the similarity, the similarity with the term vector A
Maximum word classification is denoted as g, and the cluster centre vector is the mean vector per all term vectors in class word；If d meets
The corresponding words of the term vector A, then be referred in classification g by predetermined threshold；Otherwise, create a classification h, and by institute's predicate to
The corresponding words of amount A are referred in the classification h.
Alternatively, the calculation formula of the burstiness is：
Wherein,It is burstiness, t is current time,It is the speed at current time, Δ T1、ΔT2
It is the size for the time window chosen.
Alternatively, the calculation formula of the speed at the current time is：
Wherein,It is the speed at current time, XiIt is the number that current search word occurs in i-th of search instruction,
tiIt is the timestamp of i-th of search instruction, Δ T is the size for the time window chosen.
A kind of another aspect according to embodiments of the present invention, there is provided electronic equipment of anticipation trend word.
A kind of electronic equipment of anticipation trend word, including：One or more processors；Storage device, for storing one
Or multiple programs, when one or more of programs are performed by one or more of processors so that one or more of
The method that processor realizes the anticipation trend word that the embodiment of the present invention is provided.
A kind of another further aspect according to embodiments of the present invention, there is provided computer-readable medium.
A kind of computer-readable medium, is stored thereon with computer program, this is realized when described program is executed by processor
The method for the anticipation trend word that inventive embodiments are provided.
One embodiment in foregoing invention has the following advantages that or beneficial effect：By determining the burstiness of search term,
And the change that can quantify different search terms with determination trend word is screened to the burstiness of different search terms according to pre-defined rule
Change trend degree, and value of the variation tendency of burstiness independent of time window, so as to improve trend word Forecasting Methodology
Accuracy and flexibility.In addition, in order to judge the Hot Contents of user's concern exactly according to trend word, the present invention also passes through
Search term is clustered, the semantic word with high similarity is polymerized to one kind, and use the classification of different classes of word
Identify to replace search term, to calculate the burstiness of each classification word, so that more preferable according to the burstiness of different classes of word
The Hot Contents of ground analysis user's concern, and then the user experience of network service platform is improved, brought for network service platform
More incomes.
Further effect adds hereinafter in conjunction with embodiment possessed by above-mentioned non-usual optional mode
With explanation.
Brief description of the drawings
Attached drawing is used to more fully understand the present invention, does not form inappropriate limitation of the present invention.Wherein：
Fig. 1 is the schematic diagram of the main flow of the method for anticipation trend word according to embodiments of the present invention；
Fig. 2 is the velocity function curve map of the embodiment of the present invention；
Fig. 3 is the schematic diagram of the main modular of the device of anticipation trend word according to embodiments of the present invention；
Fig. 4 is that the embodiment of the present invention can be applied to exemplary system architecture figure therein；
Fig. 5 is adapted for the structural representation for realizing the terminal device of the embodiment of the present invention or the computer system of server
Figure.
Embodiment
Explain below in conjunction with attached drawing to the one exemplary embodiment of the present invention, including the various of the embodiment of the present invention
Details should think them only exemplary to help understanding.Therefore, those of ordinary skill in the art should recognize
Arrive, various changes and modifications can be made to the embodiments described herein, without departing from scope and spirit of the present invention.Together
Sample, for clarity and conciseness, eliminates the description to known function and structure in following description.
In order to solve the problems, such as to mention in the prior art, the present invention provides a kind of method of anticipation trend word, by true
Determine the burstiness of search term, and the burstiness of different search terms is screened with determination trend word according to pre-defined rule.This hair
The method of bright anticipation trend word can be by the variation tendency degree of the burst metrization difference search term of search term, and burstiness
Value of the variation tendency independent of time window, so as to improve accuracy and the flexibility of trend word Forecasting Methodology.Separately
Outside, it is of the invention also by being clustered to search term in order to judge the Hot Contents of user's concern exactly according to trend word, with
The semantic word with high similarity is polymerized to one kind, and search term is replaced using the classification logotype of different classes of word, with
The burstiness of each classification word is calculated, so as to preferably analyze the hot spot of user's concern according to the burstiness of different classes of word
Content.
Technical scheme and its implementation process are introduced with reference to specific embodiment.
Fig. 1 is the schematic diagram of the main flow of the method for anticipation trend word according to embodiments of the present invention.As shown in Figure 1,
The method of anticipation trend word according to embodiments of the present invention mainly includes steps S101 and step S102.
Step S101：Determine the burstiness of the search term in predetermined amount of time.
Wherein, before the burstiness for determining the search term in predetermined amount of time, predetermined amount of time can also be obtained in advance
Interior search term, to calculate the burstiness of the search term of acquisition.The variation tendency of recent search term in order to obtain, the scheduled time
Section is such as can be nearest 15 days or one month.Search term can for example be obtained by the log recording data of website.It is logical
Search instruction and the corresponding time stamp data collected in nearest a period of time are crossed, and determines the search that search instruction includes
Word, you can obtain the search term in this time and corresponding time stamp data.
Under normal conditions, if some search term has obvious ascendant trend, you can it is higher to think that the search term has
Burstiness, using the search term as Bursty Words (burst word), also just obtained trend word mentioned in the present invention.
Since paroxysmal topic is typically to be triggered by some events, such as some breaking news or noticeable basket
Ball match etc., therefore this kind of topic has obtained the highest attention of user, so as to cause user to search for related content in large quantities.Search
The degree of word burst can similarly be considered " power " in physics, and in physics, " power " can use " acceleration " to represent,
" acceleration " describes in the unit interval change rate of " speed ", and the speed of " speed " expression object of which movement, i.e. search term are searched
The increased speed of rope amount.Burst word has significant acceleration in burst, and it is zero acceleration that general word, which is usually approximately considered,.
Therefore, burstiness proposed by the present invention is to determine burst word by using for reference the concept of " acceleration ", and filters out others
Non-burst search word information.
According to an embodiment of the invention, the calculation formula of burstiness for example can be：
Wherein,It is burstiness, t is current time,It is the speed at current time, Δ T1、ΔT2
It is the size for the time window chosen.
Also, the calculation formula of the speed at the current time occurred in the calculation formula of burstiness for example can be：
Wherein,It is the speed at current time, XiIt is the number that current search word occurs in i-th of search instruction,
tiIt is the timestamp of i-th of search instruction, Δ T is the size for the time window chosen.
SpeedIn exponential part exp ((ti- t)/Δ T) moving window is similar to, can be by the search of user
In word, the quantity of the search term nearer apart from current time is multiplied by greater weight, the search term of current time of adjusting the distance farther out
Quantity is multiplied by less weight, and rational time decay is carried out with the burst severity for the search term of current time farther out of adjusting the distance,
So as to reasonably be expanded to the burst severity of current time search term.Smoothing parameter Δ T is the size of traveling time window.
For the change rate of calculating speed, burstiness is defined as different time window size Δ T1With Δ T2Speed difference.
SpeedCan essentially be approximately considered be Δ T function, its function curve is as shown in Figure 2.Fig. 2 is this hair
The velocity function curve map of bright embodiment, wherein, functional value F (x) refers to speedVariable x refers to Δ T.By Fig. 2
As can be seen that the function is in x>It is decreasing function on 1 section, therefore in burstinessFormula in, for different time windows
Mouth Δ T1With Δ T2, the corresponding molecule of each time window is front and rear opposite with the position of denominator, can so ensure burstinessIt is worth for positive number.
In addition, according to burstinessCalculation formula understand, time window Δ T1With Δ T2Value before unequal
Put, no matter Δ T1With Δ T2Value is how many, and the overall variation trend for the burstiness being calculated is constant, and change is only to become
The amplitude size of change trend.Therefore, the burstiness of search term is calculated using the calculation formula of burstiness provided by the present invention,
Solve the problems, such as that burstiness need to rely on time window in the prior art.
Pass through above-mentioned computational methods, you can determine the burstiness of each search term.
Step S102：Burstiness is met that the search term of pre-defined rule is determined as trend word.
In determination trend word, can be performed according to pre-defined rule.Pre-defined rule is, for example,：Burstiness is subjected to descending
Arrangement, then obtains the search term of the highest predetermined number of burstiness (or predetermined ratio) as trend word；Or will burst
Degree meets the search term of default burstiness threshold requirement as trend word, etc..The rule of determination trend word can be as needed
Flexibly set to meet different scene needs.
According to above-mentioned step S101 and step S102, you can quantify the variation tendency degree of different search terms, and happen suddenly
Value of the variation tendency of degree independent of time window, so as to improve accuracy and the flexibility of trend word Forecasting Methodology.
In addition, according to another embodiment of the invention, in order to judge the hot spot of user's concern exactly according to trend word
Content, the present invention are polymerized to one kind also by being clustered to search term, by the semantic word with high similarity, and using not
The classification logotype of generic word replaces search term, to calculate the burstiness of each classification word, so that according to different classes of
The burstiness of word preferably analyzes the Hot Contents of user's concern.
Technical solution in accordance with another embodiment of the present invention, can also be first by semantic similarity before step S101
Meet that the word of predetermined threshold gathers for one kind, and determine the classification logotype per class word；Then by the search in predetermined amount of time
Word replaces with classification logotype.In this way, the burst per class word can be calculated according to previously described step S101 and step S102
Degree, and burstiness is met that a kind of word of pre-defined rule is determined as trend word, so as to preferably be divided according to definite trend word
Analyse the Hot Contents of user's concern.
Wherein, semantic similarity is met that the word of predetermined threshold gathers for a kind of step, it is specific hold implementation when, can be with
Performed according to the steps：
Step S1001：Obtain corpus data；
Step S1002：Participle operation is carried out to corpus data；
Step S1003：The term vector of each word obtained after generation participle operation；
Step S1004：Semantic similarity is met that the word of predetermined threshold gathers for one kind according to term vector.
Wherein, step S1004 successively can perform each term vector following operation in specific perform, until by institute
The corresponding word of some term vectors is referred to corresponding classification, and the operation of execution includes：
Obtain term vector A；
The similarity of the term vector A and the cluster centre vector per class word are calculated, by the maximum in the similarity
Value is denoted as d, is denoted as g with the word classification of the similarity maximum of the term vector A, the cluster centre vector is per in class word
The mean vector of all term vectors；
If d meets predetermined threshold, the corresponding words of the term vector A are referred in classification g；Otherwise, a class is created
Other h, and the corresponding words of the term vector A are referred in the classification h.
, can be by site information provider, search engine service provider or electric business platform etc. when obtaining corpus data
All search data in the certain period of time preserved in the database of network service platform, can also be by respectively as corpus data
The execution journal of network service platform nearest a period of time is as corpus data, etc..Corpus data, can be according to not when choosing
Property and feature with network service platform carry out the selection being directed to., can be by existing electric business platform by taking electric business platform as an example
Under all commodity description as corpus data.
Below by taking electric business platform as an example, the implementation process of another embodiment of the present invention is introduced.
The implementation of an alternative embodiment of the invention is：All descriptive labellings under existing electric business platform are collected first to make
For the corpus data of training, Chinese word segmentation instrument is recycled to segment corpus data, the word obtained after participle is operated
As the input data of the kit word2vec models for generating term vector, generate each word multidimensional (such as：200 dimensions)
Vector, i.e., each word corresponds to a multidimensional real vector in semantic space, and then can carry out word cluster；Use the present invention
In the clustering algorithm that is previously mentioned, term vector is clustered, the high word of semantic similarity can be polymerized to one kind；Then, will receive
The search term of electric business platform replaces with the classification logotype of the word classification where the word in a period of time of collection, and according to record
The timestamp of each search term, and the calculation formula of burstiness proposed by the present invention, calculate different terms classification when current
The burstiness at quarter, so as to obtain the corresponding word classification of the highest search term of burstiness, can find trend near real-time
The word classification of word, so that the hot information of user's concern is more accurately obtained, except the load for guidance search server
Equilibrium, can also instruct advertisement serving policy, the prediction sales volume of the product to and guide purchase quantity of the Cai Xiao departments to some products
Deng behavior, user experience is improved.
With certain descriptive labelling information under the electric business platform of collection, " 2017 summers of Adidas ADIDAS man trains
Exemplified by serial short-sleeve T-shirt S98731L codes ".After this descriptive labelling information is collected into as corpus data, Chinese point will be utilized
Word instrument carries out participle operation to corpus data.Common Chinese word segmentation instrument is for example stammered Chinese word segmentation, Chinese lexical analysis
System ICTCLAS (Institute of Computing Technology, Chinese Lexical Analysis System
Abbreviation), simple Chinese automatic word-cut SCWS (abbreviation of Simple Chinese Words Segmentation), etc.,
It can realize the participle function of the present invention.Multiple words will be obtained after participle operation, it is assumed that descriptive labelling information as above is divided
The word obtained after word is：" Adidas ", " ADIDAS ", " 2017 summer ", " man ", " training ", " series ", " cotta ", " T
Sympathize ", " S98731 ", " L codes ".
Afterwards, the kit word2vec for obtaining term vector is input to by each word obtained after participle is operated
In, you can the multidimensional term vector of each word is obtained, specific dimension can be set as needed, and dimension is more, calculate word phase
Result like degree is more accurate, but correspondingly computational efficiency can decrease.Word2vec uses Distributed
The term vector representation of Representation (distribution characterization), it is a by word to be that Google increases income in year in 2013
The efficient tool of real number value vector is characterized as, it utilizes the thought of deep learning, can be by the training to language material, to text
The processing of word is reduced to the vector operation in n-dimensional vector space, and the similarity in vector space can be used for representing word language
Similarity in justice.
After obtaining the term vector of each word, it is by semantic similarity being met, the word of predetermined threshold gathers according to term vector
It is a kind of., will, it is necessary to pre-set a similarity threshold Th when being clustered according to the corresponding multi-C vector of each word to word
Two words that similarity is more than the threshold value gather for one kind.
The calculation of similarity can use the cosine similarity between term vector between word and word.Assuming that term vector
Dimension is n, term vector A=(A1, A2 ..., An), term vector B='s (B1, B2 ..., Bn), then term vector A and term vector B
Cosine similarity is：
The detailed process of the clustering algorithm of the present invention is as follows：
(1) current class set G is initialized as sky, rule of thumb set similarity threshold (such as：Th=is set
0.8)；
(2) each term vector in scan data set successively；
(3) term vector is obtained；
(4) if the end of scan, algorithm terminate the term vector in data acquisition system；Otherwise step (5) is performed；
(5) distance of current term vector and each categorical clusters center vector in current class set G is calculated, will wherein most
Big similarity is denoted as d, and corresponding classification is denoted as g；
(6) if d>=Th and G are not sky, then execution step (7)；Otherwise step (8) is performed；
(7) currentElement is referred to classification g, and the more cluster centre of new category g；Return to step (3)；
(8) a classification h is created, currentElement is referred to classification h, and the more cluster centre of new category h, by the category
It is included into category set G；Return to step (3).
In above-mentioned clustering algorithm, the cluster centre vector of word classification is the average per all term vectors in class word
Vector.That is,：For certain one-dimensional vector of cluster centre vector, its vector value is the dimensional vector of all term vectors in the category
The average of value.
According to above-mentioned clustering algorithm, the cluster of the high word of semantic similarity can be completed, such as by " A Dida
This ", " ADIDAS " gather for one kind, " ice cream ", " ice cream ", " ice lolly ", " ice cream " are gathered for one kind, by " man ", " man
Money ", " man " gather for one kind, and " sport footwear ", " running shoe " are gathered for one kind etc., gather the high word of similarity for one so as to reach
The purpose of class, subsequently to show that more relevant commodity lay the first stone according to search term.
After being clustered to word, to each word classification it needs to be determined that classification logotype, for representing the word class
All words under not.Classification logotype can occur the highest word of word frequency or same word classification in the word classification
In any one word.Preserved it is then possible to which the information such as word and classification logotype that different word classifications is included corresponds to,
For use in the word classification of the method determination trend word of anticipation trend word according to the present invention.
When needing anticipation trend word, the search term of electric business platform in nearest a period of time of collection is replaced with into the word institute
Word classification classification logotype, then according to previously described step S101 and step S102, pass through each of record and search
The timestamp of rope word, and the calculation formula of burstiness proposed by the present invention, calculate different terms classification dashing forward at current time
Hair degree, so as to obtain the corresponding word classification of the highest search term of burstiness, so as to preferably analyze the focus of user.
For Network information provider, anticipation trend word can help it more targetedly to emphasize and optimize in its delivery
Hot spot part in appearance, so as to attract more flowing of access, is greatly benefited；For search engine service provider, prediction
Upcoming network trends, can help Optimizing Search as a result, at the same time by the rearrangement mode of having time dependence
The equilibrium assignmen that coming big flow keyword carries out server load can be directed to；For electric business platform, except for referring to
The load balancing of search server is led, advertisement serving policy, the prediction sales volume of the product can also be instructed to and guide Cai Xiao departments pair
The behaviors such as the purchase quantity of some products, lift user experience.
Fig. 3 is the schematic diagram of the main modular of the device of anticipation trend word according to embodiments of the present invention.As shown in figure 3,
The device 300 that the present invention applies the anticipation trend word of example mainly includes computing module 301 and determining module 302.
Computing module 301 is used for the burstiness for determining the search term in predetermined amount of time；
Determining module 302 is used to burstiness meeting that the search term of pre-defined rule is determined as trend word.
According to an embodiment of the invention, the device 300 of anticipation trend word can also include cluster module (not shown),
For：Before the burstiness for determining the search term in predetermined amount of time, the word that semantic similarity meets predetermined threshold is gathered
For one kind, and determine the classification logotype per class word；Search term in predetermined amount of time is replaced with into classification logotype.
According to an embodiment of the invention, cluster module can be also used for：
Obtain corpus data；
Participle operation is carried out to corpus data；
The term vector of each word obtained after generation participle operation；
Semantic similarity is met that the word of predetermined threshold gathers for one kind according to term vector.
In addition, cluster module can be also used for：Following operation is performed to each term vector successively, until by all words
The corresponding word of vector is referred to corresponding classification, and operation includes：
Obtain term vector A；
The similarity of term vector A and the cluster centre vector per class word are calculated, the maximum in similarity is denoted as d,
G is denoted as with the word classification of the similarity maximum of term vector A, cluster centre vector is the average per all term vectors in class word
Vector；
If d meets predetermined threshold, the corresponding words of term vector A are referred in classification g；Otherwise, a classification h is created,
And the corresponding words of term vector A are referred in classification h.
Technical solution according to embodiments of the present invention, the calculation formula of burstiness for example can be：
Wherein,It is burstiness, t is current time,It is the speed at current time, Δ T1、ΔT2
It is the size for the time window chosen.
In the calculation formula of above-mentioned burstiness, the calculation formula of the speed at current time is, for example,：
Wherein,It is the speed at current time, XiIt is the number that current search word occurs in i-th of search instruction,
tiIt is the timestamp of i-th of search instruction, Δ T is the size for the time window chosen.
Technical solution according to embodiments of the present invention, by determining the burstiness of search term, and according to pre-defined rule to not
Burstiness with search term is screened and can be quantified the variation tendency degree of different search terms with determination trend word, and is happened suddenly
Value of the variation tendency of degree independent of time window, so as to improve accuracy and the flexibility of trend word Forecasting Methodology.
In addition, in order to judge the Hot Contents of user's concern exactly according to trend word, it is of the invention also by being clustered to search term,
The semantic word with high similarity is polymerized to one kind, and search term is replaced using the classification logotype of different classes of word,
To calculate the burstiness of each classification word, so as to preferably analyze the heat of user's concern according to the burstiness of different classes of word
Point content, and then the user experience of network service platform is improved, bring more incomes for network service platform.
For Network information provider, anticipation trend word can help it more targetedly to emphasize and optimize in its delivery
Hot spot part in appearance, so as to attract more flowing of access, is greatly benefited；For search engine service provider, prediction
Upcoming network trends, can help Optimizing Search as a result, at the same time by the rearrangement mode of having time dependence
The equilibrium assignmen that coming big flow keyword carries out server load can be directed to；For electric business platform, except for referring to
The load balancing of search server is led, advertisement serving policy, the prediction sales volume of the product can also be instructed to and guide Cai Xiao departments pair
The behaviors such as the purchase quantity of some products, lift user experience.
Fig. 4 show can apply the embodiment of the present invention anticipation trend word method or anticipation trend word device it is exemplary
System architecture 400.
As shown in figure 4, system architecture 400 can include terminal device 401,402,403, network 404 and server 405.
Network 404 between terminal device 401,402,403 and server 405 provide communication link medium.Network 404 can be with
Including various connection types, such as wired, wireless communication link or fiber optic cables etc..
User can be interacted with using terminal equipment 401,402,403 by network 404 with server 405, to receive or send out
Send message etc..Various telecommunication customer end applications, such as the application of shopping class, net can be installed on terminal device 401,402,403
(merely illustrative) such as the application of page browsing device, searching class application, instant messaging tools, mailbox client, social platform softwares.
Terminal device 401,402,403 can have a display screen and a various electronic equipments that supported web page browses, bag
Include but be not limited to smart mobile phone, tablet computer, pocket computer on knee and desktop computer etc..
Server 405 can be to provide the server of various services, such as utilize terminal device 401,402,403 to user
The shopping class website browsed provides the back-stage management server (merely illustrative) supported.Back-stage management server can be to receiving
To the data such as information query request analyze etc. processing, and by handling result (such as target push information, product letter
Breath -- merely illustrative) feed back to terminal device.
It should be noted that the anticipation trend word method that the embodiment of the present invention is provided generally is performed by server 405, phase
Ying Di, anticipation trend word device are generally positioned in server 405.
It should be understood that the number of the terminal device, network and server in Fig. 4 is only schematical.According to realizing need
Will, can have any number of terminal device, network and server.
Below with reference to Fig. 5, it illustrates suitable for for realizing the computer system 500 of the electronic equipment of the embodiment of the present invention
Structure diagram.Terminal device shown in Fig. 5 is only an example, to the function of the embodiment of the present invention and should not use model
Shroud carrys out any restrictions.
As shown in figure 5, computer system 500 includes central processing unit (CPU) 501, it can be read-only according to being stored in
Program in memory (ROM) 502 or be loaded into program in random access storage device (RAM) 503 from storage part 508 and
Perform various appropriate actions and processing.In RAM 503, also it is stored with system 500 and operates required various programs and data.
CPU 501, ROM 502 and RAM 503 are connected with each other by bus 504.Input/output (I/O) interface 505 is also connected to always
Line 504.
I/O interfaces 505 are connected to lower component：Importation 506 including keyboard, mouse etc.；Penetrated including such as cathode
The output par, c 507 of spool (CRT), liquid crystal display (LCD) etc. and loudspeaker etc.；Storage part 508 including hard disk etc.；
And the communications portion 509 of the network interface card including LAN card, modem etc..Communications portion 509 via such as because
The network of spy's net performs communication process.Driver 510 is also according to needing to be connected to I/O interfaces 505.Detachable media 511, such as
Disk, CD, magneto-optic disk, semiconductor memory etc., are installed on driver 510, in order to read from it as needed
Computer program be mounted into as needed storage part 508.
Especially, disclosed embodiment, the process described above with reference to flow chart may be implemented as counting according to the present invention
Calculation machine software program.For example, embodiment disclosed by the invention includes a kind of computer program product, it includes being carried on computer
Computer program on computer-readable recording medium, the computer program include the program code for being used for the method shown in execution flow chart.
In such embodiment, which can be downloaded and installed by communications portion 509 from network, and/or from can
Medium 511 is dismantled to be mounted.When the computer program is performed by central processing unit (CPU) 501, system of the invention is performed
The above-mentioned function of middle restriction.
It should be noted that the computer-readable medium shown in the present invention can be computer-readable signal media or meter
Calculation machine readable storage medium storing program for executing either the two any combination.Computer-readable recording medium for example can be --- but not
Be limited to --- electricity, magnetic, optical, electromagnetic, system, device or the device of infrared ray or semiconductor, or it is any more than combination.Meter
The more specifically example of calculation machine readable storage medium storing program for executing can include but is not limited to：Electrical connection with one or more conducting wires, just
Take formula computer disk, hard disk, random access storage device (RAM), read-only storage (ROM), erasable type and may be programmed read-only storage
Device (EPROM or flash memory), optical fiber, portable compact disc read-only storage (CD-ROM), light storage device, magnetic memory device,
Or above-mentioned any appropriate combination.In the present invention, computer-readable recording medium can any include or store journey
The tangible medium of sequence, the program can be commanded the either device use or in connection of execution system, device.And at this
In invention, computer-readable signal media can include in a base band or as carrier wave a part propagation data-signal,
Wherein carry computer-readable program code.The data-signal of this propagation can take various forms, including but unlimited
In electromagnetic signal, optical signal or above-mentioned any appropriate combination.Computer-readable signal media can also be that computer can
Any computer-readable medium beyond storage medium is read, which, which can send, propagates or transmit, is used for
By instruction execution system, device either device use or program in connection.Included on computer-readable medium
Program code can be transmitted with any appropriate medium, be included but not limited to：Wirelessly, electric wire, optical cable, RF etc., or it is above-mentioned
Any appropriate combination.
Flow chart and block diagram in attached drawing, it is illustrated that according to the system of various embodiments of the invention, method and computer journey
Architectural framework in the cards, function and the operation of sequence product.At this point, each square frame in flow chart or block diagram can generation
The part of one module of table, program segment or code, a part for above-mentioned module, program segment or code include one or more
The executable instruction of logic function as defined in being used for realization.It should also be noted that some as replace realization in, institute in square frame
The function of mark can also be with different from the order marked in attached drawing generation.For example, two square frames succeedingly represented are actual
On can perform substantially in parallel, they can also be performed in the opposite order sometimes, this is depending on involved function.Also
It is noted that the combination of each square frame and block diagram in block diagram or flow chart or the square frame in flow chart, can use and perform rule
The dedicated hardware based systems of fixed functions or operations is realized, or can use the group of specialized hardware and computer instruction
Close to realize.
Being described in unit or module involved in the embodiment of the present invention can be realized by way of software, can also
Realized by way of hardware.Described unit or module can also be set within a processor, for example, can be described as：
A kind of processor includes computing module and determining module.Wherein, the title of these units or module not structure under certain conditions
The paired restriction of the unit or module in itself, for example, computing module is also described as " being used to determine in predetermined amount of time
The module of the burstiness of search term ".
As on the other hand, present invention also offers a kind of computer-readable medium, which can be
Included in equipment described in above-described embodiment；Can also be individualism, and without be incorporated the equipment in.Above-mentioned calculating
Machine computer-readable recording medium carries one or more program, when said one or multiple programs are performed by the equipment, makes
Obtaining the equipment includes：Determine the burstiness of the search term in predetermined amount of time；Search term that burstiness meets pre-defined rule is true
It is set to trend word.
Technical solution according to embodiments of the present invention, by determining the burstiness of search term, and according to pre-defined rule to not
Burstiness with search term is screened and can be quantified the variation tendency degree of different search terms with determination trend word, and is happened suddenly
Value of the variation tendency of degree independent of time window, so as to improve accuracy and the flexibility of trend word Forecasting Methodology.
In addition, in order to judge the Hot Contents of user's concern exactly according to trend word, it is of the invention also by being clustered to search term,
The semantic word with high similarity is polymerized to one kind, and search term is replaced using the classification logotype of different classes of word,
To calculate the burstiness of each classification word, so as to preferably analyze the heat of user's concern according to the burstiness of different classes of word
Point content, and then the user experience of network service platform is improved, bring more incomes for network service platform.
Above-mentioned embodiment, does not form limiting the scope of the invention.Those skilled in the art should be bright
It is white, depending on design requirement and other factors, various modifications, combination, sub-portfolio and replacement can occur.It is any
Modifications, equivalent substitutions and improvements made within the spirit and principles in the present invention etc., should be included in the scope of the present invention
Within.
Claims (14)
- A kind of 1. method of anticipation trend word, it is characterised in that including：Determine the burstiness of the search term in predetermined amount of time；The described search word that the burstiness meets pre-defined rule is determined as trend word.
- 2. according to the method described in claim 1, it is characterized in that, determine the step of the burstiness of the search term in predetermined amount of time Before rapid, further include：Semantic similarity is met that the word of predetermined threshold gathers for one kind, and determines the classification logotype per class word；Search term in predetermined amount of time is replaced with into the classification logotype.
- 3. according to the method described in claim 2, it is characterized in that, semantic similarity is met that the word of predetermined threshold gathers for one The step of class, includes：Obtain corpus data；Participle operation is carried out to the corpus data；Generate the term vector of each word obtained after the participle operation；Semantic similarity is met that the word of predetermined threshold gathers for one kind according to the term vector.
- 4. according to the method described in claim 3, it is characterized in that, semantic similarity is met by predetermined threshold according to the term vector The word of value gathers to be included for a kind of step：Following operation is performed to each term vector successively, until all corresponding words of term vector are referred to corresponding class Not, the operation includes：Obtain term vector A；The similarity of the term vector A and the cluster centre vector per class word are calculated, the maximum in the similarity is remembered Make d, be denoted as g with the word classification of the similarity maximum of the term vector A, the cluster centre vector is to own per in class word The mean vector of term vector；If d meets predetermined threshold, the corresponding words of the term vector A are referred in classification g；Otherwise, a classification h is created, And the corresponding words of the term vector A are referred in the classification h.
- 5. according to the method described in claim 1, it is characterized in that, the calculation formula of the burstiness is：Wherein,It is burstiness, t is current time,It is the speed at current time, Δ T1、ΔT2It is choosing The size of the time window taken.
- 6. according to the method described in claim 5, it is characterized in that, the calculation formula of the speed at the current time is：Wherein,It is the speed at current time, XiIt is the number that current search word occurs in i-th of search instruction, tiIt is The timestamp of i-th of search instruction, Δ T are the sizes for the time window chosen.
- A kind of 7. device of anticipation trend word, it is characterised in that including：Computing module, for determining the burstiness of the search term in predetermined amount of time；Determining module, the described search word for the burstiness to be met to pre-defined rule are determined as trend word.
- 8. device according to claim 7, it is characterised in that further include cluster module, be used for：In definite predetermined amount of time Before the burstiness of interior search term,Semantic similarity is met that the word of predetermined threshold gathers for one kind, and determines the classification logotype per class word；Search term in predetermined amount of time is replaced with into the classification logotype.
- 9. device according to claim 8, it is characterised in that the cluster module is additionally operable to：Obtain corpus data；Participle operation is carried out to the corpus data；Generate the term vector of each word obtained after the participle operation；Semantic similarity is met that the word of predetermined threshold gathers for one kind according to the term vector.
- 10. device according to claim 9, it is characterised in that the cluster module is additionally operable to：Following operation is performed to each term vector successively, until all corresponding words of term vector are referred to corresponding class Not, the operation includes：Obtain term vector A；The similarity of the term vector A and the cluster centre vector per class word are calculated, the maximum in the similarity is remembered Make d, be denoted as g with the word classification of the similarity maximum of the term vector A, the cluster centre vector is to own per in class word The mean vector of term vector；If d meets predetermined threshold, the corresponding words of the term vector A are referred in classification g；Otherwise, a classification h is created, And the corresponding words of the term vector A are referred in the classification h.
- 11. device according to claim 7, it is characterised in that the calculation formula of the burstiness is：Wherein,It is burstiness, t is current time,It is the speed at current time, Δ T1、ΔT2It is choosing The size of the time window taken.
- 12. according to the devices described in claim 11, it is characterised in that the calculation formula of the speed at the current time is：Wherein,It is the speed at current time, XiIt is the number that current search word occurs in i-th of search instruction, tiIt is The timestamp of i-th of search instruction, Δ T are the sizes for the time window chosen.
- A kind of 13. electronic equipment of anticipation trend word, it is characterised in that including：One or more processors；Storage device, for storing one or more programs,When one or more of programs are performed by one or more of processors so that one or more of processors are real The now method as described in any in claim 1-6.
- 14. a kind of computer-readable medium, is stored thereon with computer program, it is characterised in that described program is held by processor The method as described in any in claim 1-6 is realized during row.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710969459.6A CN107908616B (en) | 2017-10-18 | 2017-10-18 | Method and device for predicting trend words |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710969459.6A CN107908616B (en) | 2017-10-18 | 2017-10-18 | Method and device for predicting trend words |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107908616A true CN107908616A (en) | 2018-04-13 |
| CN107908616B CN107908616B (en) | 2022-01-28 |
Family
ID=61841422
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710969459.6A Active CN107908616B (en) | 2017-10-18 | 2017-10-18 | Method and device for predicting trend words |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107908616B (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110377916A (en) * | 2018-08-17 | 2019-10-25 | 腾讯科技（深圳）有限公司 | Word prediction technique, device, computer equipment and storage medium |
| CN110489741A (en) * | 2019-07-12 | 2019-11-22 | 北京邮电大学 | Microblogging burst topic detecting method based on the detection of burst word and filtering |
| CN110968691A (en) * | 2018-09-30 | 2020-04-07 | 北京国双科技有限公司 | Judicial hotspot determination method and device |
| US20210034689A1 (en) * | 2019-07-30 | 2021-02-04 | Baidu Online Network Technology (Beijing) Co., Ltd. | Method and apparatus for detecting regional event based on search engine, and storage medium |
| CN113743973A (en) * | 2020-11-30 | 2021-12-03 | 北京沃东天骏信息技术有限公司 | Method and device for analyzing market hotspot trend |
| CN117473144A (en) * | 2023-12-27 | 2024-01-30 | 深圳市活力天汇科技股份有限公司 | Method for storing route data, computer equipment and readable storage medium |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8108407B2 (en) * | 2006-11-06 | 2012-01-31 | Panasonic Corporation | Informationn retrieval apparatus |
| CN102420747A (en) * | 2011-11-17 | 2012-04-18 | 清华大学 | Service source shaping method based on date packet achieving interval wave filtration |
| CN102999539A (en) * | 2011-09-13 | 2013-03-27 | 富士通株式会社 | Method and device for forecasting future development trend of given topic |
| CN103164540A (en) * | 2013-04-15 | 2013-06-19 | 武汉大学 | Patent hotspot discovery and trend analysis method |
| CN104035960A (en) * | 2014-05-08 | 2014-09-10 | 东莞市巨细信息科技有限公司 | Internet information hotspot predicting method |
| CN104573031A (en) * | 2015-01-14 | 2015-04-29 | 哈尔滨工业大学深圳研究生院 | Micro blog emergency detection method |
-
2017
- 2017-10-18 CN CN201710969459.6A patent/CN107908616B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8108407B2 (en) * | 2006-11-06 | 2012-01-31 | Panasonic Corporation | Informationn retrieval apparatus |
| CN102999539A (en) * | 2011-09-13 | 2013-03-27 | 富士通株式会社 | Method and device for forecasting future development trend of given topic |
| CN102420747A (en) * | 2011-11-17 | 2012-04-18 | 清华大学 | Service source shaping method based on date packet achieving interval wave filtration |
| CN103164540A (en) * | 2013-04-15 | 2013-06-19 | 武汉大学 | Patent hotspot discovery and trend analysis method |
| CN104035960A (en) * | 2014-05-08 | 2014-09-10 | 东莞市巨细信息科技有限公司 | Internet information hotspot predicting method |
| CN104573031A (en) * | 2015-01-14 | 2015-04-29 | 哈尔滨工业大学深圳研究生院 | Micro blog emergency detection method |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110377916A (en) * | 2018-08-17 | 2019-10-25 | 腾讯科技（深圳）有限公司 | Word prediction technique, device, computer equipment and storage medium |
| CN110377916B (en) * | 2018-08-17 | 2022-12-16 | 腾讯科技（深圳）有限公司 | Word prediction method, word prediction device, computer equipment and storage medium |
| CN110968691A (en) * | 2018-09-30 | 2020-04-07 | 北京国双科技有限公司 | Judicial hotspot determination method and device |
| CN110489741A (en) * | 2019-07-12 | 2019-11-22 | 北京邮电大学 | Microblogging burst topic detecting method based on the detection of burst word and filtering |
| CN110489741B (en) * | 2019-07-12 | 2022-06-21 | 北京邮电大学 | Microblog burst topic detection method based on burst word detection and filtering |
| US20210034689A1 (en) * | 2019-07-30 | 2021-02-04 | Baidu Online Network Technology (Beijing) Co., Ltd. | Method and apparatus for detecting regional event based on search engine, and storage medium |
| US11449567B2 (en) * | 2019-07-30 | 2022-09-20 | Baidu Online Network Technology (Beijing) Co., Ltd. | Method and apparatus for detecting regional event based on search engine, and storage medium |
| CN113743973A (en) * | 2020-11-30 | 2021-12-03 | 北京沃东天骏信息技术有限公司 | Method and device for analyzing market hotspot trend |
| CN113743973B (en) * | 2020-11-30 | 2024-07-16 | 北京沃东天骏信息技术有限公司 | Method and device for analyzing market hotspot trend |
| CN117473144A (en) * | 2023-12-27 | 2024-01-30 | 深圳市活力天汇科技股份有限公司 | Method for storing route data, computer equipment and readable storage medium |
| CN117473144B (en) * | 2023-12-27 | 2024-03-29 | 深圳市活力天汇科技股份有限公司 | Method for storing route data, computer equipment and readable storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107908616B (en) | 2022-01-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107908616A (en) | The method and apparatus of anticipation trend word | |
| KR101939554B1 (en) | Determining a temporary transaction limit | |
| CN104254852B (en) | Method and system for mixed information inquiry | |
| CN109189904A (en) | Individuation search method and system | |
| CN109684538A (en) | A kind of recommended method and recommender system based on individual subscriber feature | |
| CN108345702A (en) | Entity recommends method and apparatus | |
| US20120254242A1 (en) | Methods and systems for mining association rules | |
| CN112148987A (en) | Message pushing method based on target object activity and related equipment | |
| US20090276729A1 (en) | Adaptive user feedback window | |
| CN107609890A (en) | A kind of method and apparatus of order tracking | |
| CN112115363A (en) | Recommendation method, computing device and storage medium | |
| CN106682686A (en) | User gender prediction method based on mobile phone Internet-surfing behavior | |
| CN110135976A (en) | User's portrait generation method, device, electronic equipment and computer-readable medium | |
| CN107451832A (en) | The method and apparatus of pushed information | |
| CN108932625B (en) | User behavior data analysis method, device, medium and electronic equipment | |
| CN110163661A (en) | Marketing message promotion method, device, electronic equipment and computer-readable medium | |
| CN111967914A (en) | User portrait based recommendation method and device, computer equipment and storage medium | |
| CN110363604A (en) | Page generation method and device | |
| US20220284067A1 (en) | Method for pushing information, electronic device | |
| CN111400613A (en) | Article recommendation method, device, medium and computer equipment | |
| CN114065750A (en) | Commodity information matching and publishing method and device, equipment, medium and product thereof | |
| CN113051480A (en) | Resource pushing method and device, electronic equipment and storage medium | |
| CN107885783A (en) | The method and apparatus for obtaining the high relevant classification of search term | |
| CN107357794A (en) | Optimize the method and apparatus of the data store organisation of key value database | |
| CN115204881A (en) | Data processing method, device, equipment and storage medium |
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 |