CN102214186A - Method and system for displaying object relation - Google Patents
Method and system for displaying object relation Download PDFInfo
- Publication number
- CN102214186A CN102214186A CN2010101460610A CN201010146061A CN102214186A CN 102214186 A CN102214186 A CN 102214186A CN 2010101460610 A CN2010101460610 A CN 2010101460610A CN 201010146061 A CN201010146061 A CN 201010146061A CN 102214186 A CN102214186 A CN 102214186A
- Authority
- CN
- China
- Prior art keywords
- classification
- information
- show
- conjunctive word
- relationship
- 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
Images
Landscapes
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention relates to a method for displaying an object relation. The method comprises the following steps of: A, acquiring objects from internet information and collecting a co-occurrence frequency of the objects; B, computing a total frequency of the objects which occur two by two in a predetermined date according to the co-occurrence frequency, computing the closeness of the object relation according to the total frequency, and obtaining the object to be displayed; C, acquiring a correlative word of the object relation to be displayed in the predetermined date from the internet information according to the object to be displayed; and D, outputting and displaying according to the object to be displayed and the correlative word of the object relation to be displayed. Moreover, the invention also relates to a system for displaying the object relation. In the method and the system for displaying the object relation, the object relation and the correlative word of the object relation can be immediately updated by computing the closeness of the object relation in the predetermined date, updating the object relation at fixed time and acquiring the correlative word of the object relation in the predetermined date, so that relations between the objects can be timely displayed, and the timeliness of the object relation is embodied.
Description
[technical field]
The present invention relates to field of computer technology, relate in particular to a kind of method and system of showing object relationship.
[background technology]
In field of computer technology, include relation between objects in the various internet informations, concern as the personage etc.The tradition searching for Internet information is also showed in the method that concerns between the personage, first method: specific personage is correlated with everyone all present, for example, search column is keyed in the name of " Yao Ming ", the design sketch that shows a two dimension at once, it is a Yao Ming's graph of a relation, shades of colour, the different name of distance appear around Yao Ming's name, click the white line that connects between two names, demonstrate the relation of these names on the figure: teammate, adversary, girl friend, successor, middle man with Yao Ming.Second method: the information branch of a radial of employing presents the relation between the personage, and as numerous feeler is arranged on the rudder for ship, each feeler all is a relevant information, when clicking a feeler, and the information branch of extending another hiding radial.The third method: by the AD HOC coupling, perhaps the human-edited extracts the static incidence relation between the personage (as relations such as father and son, lover, masters and apprentices).These three kinds of ways of presentation all are the static relations between the personage who shows, can not change because the time changes.
[summary of the invention]
Based on this, be necessary to provide a kind of method of showing object relationship, can show relation between objects more timely.
A kind of method of showing object relationship may further comprise the steps:
A. from internet information, obtain the object and the acquisition target co-occurrence frequency;
B. calculate total frequency that target date, object occurred in twos according to the described co-occurrence frequency, again according to total frequency calculating object degree in close relations and draw object to be showed;
C. according to waiting to show that object obtains treating in target date and shows the object relationship conjunctive word from internet information;
D. wait to show object and wait to show that the object relationship conjunctive word exports displaying according to described.
Preferably, the computing formula that object relationship is spent closely among the described step B is:
Wherein, R (A B) is the degree in close relations of any object A, B, and F (A, B, D) the total frequency that occurs in twos for object A, the B of date D, α and β are adjustable parameter, Distan ce (CurrentDay, D) time gap of expression date D and current date.
Preferably, step C comprises:
C1. choose the candidate association speech of waiting to show object relationship;
C2. calculate the weight of the candidate association speech wait to show object relationship;
C3. the timeliness weight of calculating the candidate association speech wait to show object relationship draws waits to show the object relationship conjunctive word in target date.
Preferably, the formula that calculates the weight of the candidate association speech wait to show object relationship among the step C2 is:
F(w，d)＝log(length(w))*P(TF(w)*IDF(w))*H(w)*Title(w)
Wherein, (d represents the date to F for w, the d) weighted value of expression conjunctive word w; Length (w) is the length of conjunctive word, and TF (w) is the word frequency of conjunctive word w in internet information, and IDF (w) is meant the contrary document frequency of speech w; Function P is a piecewise function; H (w) is regular weight; Title (w) is meant the weight that conjunctive word w increases when appearing at the title of internet information, and default value is 1.
Preferably, the method for described displaying object relationship also comprises step e: obtain wait to show object after, judge describedly wait to show whether object is the duplication of name object, if the duplication of name object is handled.
Preferably, step e further may further comprise the steps:
E1. set up duplication of name list object information;
E2. make up the classification information hierarchical tree according to the internet historical information;
E3. according to the classification information hierarchical tree that makes up the duplication of name object is handled.
Preferably, step e 2 comprises:
E21. obtain the internet historical information and from described internet historical information, extract classification information, and classification information is carried out pre-service, obtain the classification information set of strings;
E22. resolve classification information string in the described classification information set of strings, calculate the probability of any classification set membership in the described classification information string;
E23. according to the probability initialization classification information hierarchical tree of described classification information string and any classification set membership, choose the set of root classification;
E24. the arbitrary classification of recurrence expands to the path of root classification, obtain in such path that is clipped to root classification process all probability values of taking advantage of of two-layer classification set membership up and down, the probability that calculates this classification and be the subclass of root classification is clipped to the summation of probability value of taking advantage of described in all paths of root classification process for such.
Preferably, step e 3 comprises:
E30. judge describedly wait to show whether object is present in the duplication of name list object information, if, execution in step E31, otherwise finish;
E31. extract the classification information string of waiting to show object place internet information;
E32. judge whether described classification information string exists, if, execution in step E33 then, if not, execution in step E34 then;
E33. according to described classification information hierarchical tree, obtain the probability of the subclass as the root classification of all categories in the described classification information string, the root classification of getting described probability maximum is as the output result;
E34. retrieve the information distinguished of duplication of name object in the internet information, and judge whether and can distinguish, if, execution in step E35, otherwise finish;
E35. export corresponding object to be showed.
Preferably, step D specifically comprises:
D1. receive user's inputted search object;
D2. set up list object according to object search and obtain object to be showed and the object relationship conjunctive word;
D3. show according to waiting of obtaining that object and object relationship conjunctive word are treated and show that object's position carries out the initialization setting;
D4. play up according to the object's position to be showed of setpoint distance after initialization.
In addition, also be necessary to provide a kind of system that shows object relationship, can show relation between objects more timely.
A kind of system that shows object relationship, this system comprises:
Information acquisition module is used for obtaining the object and the acquisition target co-occurrence frequency from internet information;
Processing module is calculated total frequency that target date, object occurred in twos, again according to total frequency calculating object degree in close relations, and draws object to be showed;
The conjunctive word acquisition module is used for according to waiting to show that object obtains treating in target date from internet information and shows the object relationship conjunctive word;
The output display module is according to waiting to show object and waiting to show that the object relationship conjunctive word exports displaying.
Preferably, described processing module calculate described object relationship closely the formula of degree be:
Wherein, (A B) is the degree weight in close relations of object A, B to R, and (A, B D) are the object A of date D, sum frequency that B occurs in twos to F, and α and β are adjustable parameter, Dis tan ce (CurrentDay, D) time gap of expression date D and current date.
Preferably, the conjunctive word acquisition module is further used for choosing the candidate association speech of waiting to show object relationship, calculates the candidate association speech weight wait to show object relationship and conjunctive word timeliness weight and draws and wait to show the object relationship conjunctive word in target date.
Preferably, the weight formula of the candidate association speech of described conjunctive word acquisition module calculating object relation is:
F(w，d)＝log(length(w))*P(TF(w)*IDF(w))*H(w)*Title(w)
Wherein, (d represents the date to F for w, the d) weighted value of expression conjunctive word w; Length (w) is the length of conjunctive word, and TF (w) is the word frequency of conjunctive word w in internet information, and IDF (w) is meant the contrary document frequency of speech w; Function P is a piecewise function; H (w) is regular weight; Title (w) is meant the weight that conjunctive word w increases when appearing at the title of internet information, and default value is 1.
Preferably, described system also comprises duplication of name object handles module, and described duplication of name object handles module is used for after obtaining object, judges that whether object is the duplication of name object and the duplication of name object handled.
Preferably, described duplication of name object handles module is further used for setting up duplication of name list object information, makes up the classification information hierarchical tree according to historical internet historical information, according to the classification information hierarchical tree that makes up the duplication of name object is handled.
Preferably, described duplication of name object handles module further comprises:
The classification information acquiring unit extracts classification information from the internet historical information;
Pretreatment unit carries out pre-service to described classification information, obtains the classification information set of strings;
Classification information string in the classification information set of strings is resolved in the probability calculation unit, calculates the probability of any classification set membership in the described classification information string;
Classification information hierarchical tree initialization unit according to the probability initialization classification information hierarchical tree of described classification information string and any classification set membership, is chosen the set of root classification;
The recurrence expanding element, be used for the path that the arbitrary classification of recurrence expands to the root classification, obtain in such path that is clipped to root classification process all probability values of taking advantage of of two-layer classification set membership up and down, the probability that calculates this classification and be the subclass of root classification is clipped to the summation of probability value of taking advantage of described in all paths of root classification process for such.
Preferably, described duplication of name object handles module is further used for judging waits in the internet information to show that object is in duplication of name list object information, extract the classification information string of object place internet information, calculate the probability of the subclass as the root classification of all categories in the described classification information, the root classification of getting described probability maximum is as the output result.
Preferably, described output display module also comprises:
Input block is used to receive user's inputted search object;
The unit is set up in tabulation, is used for setting up list object and obtaining object to be showed and the object relationship conjunctive word according to object search;
Wait to show the object's position initialization unit, be used for showing according to waiting of obtaining that object and object relationship conjunctive word are treated and show that object's position carries out the initialization setting;
The object's position rendering unit is used for playing up according to the object's position to be showed of setpoint distance after to initialization.
The method and system of above-mentioned displaying object relationship, the object relationship that employing was calculated in target date is spent and timing upgating object relation closely, and obtain the object relationship conjunctive word in target date, upgating object concerns and the object relationship conjunctive word timely, can show relation between objects more timely like this, embody the ageing of object relationship.
In addition, the duplication of name object is distinguished judgement, improved the accuracy of object relationship; Employing is played up the position of object search, can embody object relationship clear, accurately.
[description of drawings]
Fig. 1 is a method flow diagram of showing object relationship among the embodiment;
Fig. 2 is according to waiting to show that object obtains the method flow diagram that treating in target date showed the object relationship conjunctive word from internet information among the embodiment;
Fig. 3 is the method flow diagram that output is showed among another embodiment;
Fig. 4 is to the particular flow sheet of duplication of name object handles among another embodiment;
Fig. 5 is the method flow diagram that makes up the classification information hierarchical tree among another embodiment according to the internet historical information;
Fig. 6 is according to the method flow diagram of classification information hierarchical tree to the duplication of name object handles among another embodiment;
Fig. 7 is a structural representation of showing the system of object relationship among the embodiment;
Fig. 8 is the structural representation of output display module among the embodiment;
Fig. 9 is a structural representation of showing the system of object relationship among another embodiment;
Figure 10 is the structural unit synoptic diagram of duplication of name object handles module construction information category hierarchical tree part among the embodiment.
[embodiment]
As described in Figure 1, in one embodiment, a kind of method of showing object relationship comprises the steps:
Step S10 obtains the object and the acquisition target co-occurrence frequency from internet information.Wherein, by the internet information in the internet is carried out data mining,, extract object oriented, the co-occurrence frequency and context key message according to data computation such as the classified information of internet, context of co-texts.Every news information is handled, confirmed a center object, obtain other objects that in this news, occur again, add up the frequency of object co-occurrence in many news according to headline.In the present embodiment, adopt traditional Words partition system that internet data is carried out analyzing and processing.
Step S20 calculates total frequency that target date, object occurred in twos according to the described co-occurrence frequency, again according to total frequency calculating object degree in close relations and draw object to be showed.Wherein, object relationship can be personage's relation, Brand Relationship, customer relationship, incident three dimensional interpretation aspect, internet etc.In the present embodiment, object relationship is personage's relation, then extracts person names, the statistics personage co-occurrence frequency, total frequency that the calculating personage occurs in twos and the close degree that calculates personage's relation in internet information.For example extract the central person, from the co-occurrence frequency of record, calculate total frequency that other personages occur with the central person.If it is big more that two people appear at the frequency in the internet simultaneously, it is close more to illustrate that they concern.In addition, by introducing the close degree that the time attenuation function calculates personage's relation.The time attenuation function is a variable time factor, and this time factor is
Wherein, α and β are adjustable parameter, Dis tan ce (CurrentDay, D) time gap of expression date D and current date.Then the formula of calculating object degree in close relations is:
Wherein, (A B) is the degree weight in close relations of object A, B to R, and (A, B D) are the object A of date D, sum frequency that B occurs in twos to F.For the better calculating of explanation personage degree in close relations, be exemplified below:
Two people of A and B suppose to occur the day before yesterday 4 times, occur yesterday 3 times, today occurs 5 times, if adjustable parameters α=1 so in the computing formula, then calculate according to computing formula β=2, and the weight of final A and two people's of B degree in close relations is:
Be equivalent to, occurred the day before yesterday 4 times, be the contribution of last weight
In like manner, 5 contribution margins to last comprehensive weight that occur with today for 3 times of appearance yesterday are respectively 0.75 and 5.0.
By calculating the close degree of personage's relation, the employing weighted value is represented, by comparing the size of weighted value, weighted value is sorted again, obtain the bigger personage of weighted value, personage's (object promptly to be showed) of the bigger weighted value that obtains is sent to front-end user interface.In addition, regularly upgrading the personage concerns circle, by the default time, and timing acquiring people information, the personage that upgrades in time relation.Time can be set at 5 minutes upgrade once or upgraded in 1 hour one inferior.
Step S30 shows that according to waiting object obtains treating in target date and shows the object relationship conjunctive word from internet information.By Chinese sentence part of speech combination rule, characteristic of speech sounds in the internet information being summed up the object relationship conjunctive word of the Rule Extraction sentence of associated extraction.Wherein, the object relationship conjunctive word obtains for extraction from internet information, the speech or the phrase that concern between the energy description object.As " Chen Guanxi and A Jiao overwhelm greatly in the Nude Picture Scandal incident ", in this sentence, we can extract " Nude Picture Scandal incident ", the incidence relation between personage Chen Guanxi and the A Jiao described in " overwhelming greatly " two phrases.As shown in Figure 2, the concrete step of obtaining the object relationship conjunctive word comprises:
Step S300 chooses the candidate association speech of waiting to show object relationship.In the present embodiment, judge whether contain two objects in the sentence,, then adopt traditional Words partition system that sentence in the internet information is carried out participle and part-of-speech tagging, choose the candidate association speech of waiting to show object relationship according to the Chinese sentence syntax rule if having.
Step S301 calculates the weight of the candidate association speech wait to show object relationship.Calculating waits to show that the formula of weight of candidate association speech of object relationship is as follows:
F(w，d)＝log(length(w))*P(TF(w)*IDF(w))*H(w)*Title(w)
Wherein, (d represents the date to F for w, the d) weighted value of expression conjunctive word w; Length (w) is the length of conjunctive word, has adopted the length priority principle in the present embodiment, and based on a basic hypothesis: the more semanteme of horn of plenty generally can be expressed in the Chinese phrase that number of words is many more.TF (w) is the word frequency of conjunctive word w in internet information, and IDF (w) is the information that obtains of adding up in advance, is meant the contrary document frequency of speech w.Function P is a piecewise function.H (w) is regular weight, the extraction of conjunctive word mainly depends on the Chinese sentence syntax rule, and the word that each Rule Extraction is come out can be endowed an initial weighted value, is regular weight, conjunctive word w is extracted by Different Rule and obtains, and its regular weight H value is also different.Title (w) is meant the weight that conjunctive word w increases when appearing at the title of internet information, and default value is 1.Conjunctive word that calculates and weighted value thereof deposit in the Database Systems, as the candidate word of conjunctive word between two objects that extract in the sentence.
Step S302, the timeliness weight of calculating the candidate association speech wait to show object relationship draws waits to show the object relationship conjunctive word in target date.In order to calculate the object association speech weight in target date, promptly embody the ageing of conjunctive word weight, introduce the time attenuation function, can calculate the timeliness weight of conjunctive word, be equivalent to the conjunctive word weight and multiply by a variable time factor.Formula is as follows:
Wherein, α and β are adjustable parameter, Dis tan ce (CurrentDay, d) time gap of expression date d and current date.In the present embodiment, closing with the personage is example, for two personages, can utilize following formula to calculate the timeliness weight S (w) of each conjunctive word w between two personages:
Get access to the timeliness weight of all candidates' conjunctive word between two personages, candidate's conjunctive word timeliness weight is sorted from big to small, can select several of coming the front and output to the user, as describing the conjunctive word that concerns between two personages.
Step S40 is according to waiting to show object and waiting to show that the object relationship conjunctive word exports displaying.Waiting of drawing showed object and wait to show that the object relationship conjunctive word sends to front end and shows and present.In one embodiment, as shown in Figure 3, step S40 specifically comprises:
Step S400 receives user's inputted search object.
Step S410 sets up list object and obtains object to be showed and the object relationship conjunctive word according to object search.In the present embodiment, closing with the personage is example, concern that according to the personage sort algorithm sets up personage tabulation, promptly from personage's tabulation, read a personage in order, if this person has a plurality of relations, get the next one, otherwise, this person is joined the ranking results tabulation, obtain this person's the people who concerns the other end then, search with he related first personage, join tabulation (that had searched just no longer searches).Up to all personages are joined end of list (EOL).Concrete establishment step comprises:
(1) pulls the object relationship data from server.Wherein, the object relationship data comprise center object, one-level relation data, degree relationship data.
(2) according to the object relationship data, the opening relationships tabulation.The object relationship data that circulate all, must data presented adding relation list, data presented joins interim tabulation at random, and picked at random relation data therefrom joins relation list then.Must data presented be object direct and that center object is related, according to concerning that the ordering of temperature weight size draws, data presented be indirect object relationship at random, and the data of object relationship are a lot of indirectly, can show at random.
(3) set up list object according to relation list.
Step S420 shows according to waiting of obtaining that object and object relationship conjunctive word are treated and shows that object's position carries out the initialization setting.At first the center object node is placed on the platform center, selects an angle at random, according to angle uniformly-spaced each node is placed on the position of the peripheral random distance of center object, draw and concern node according to object order result in the list object.Wherein, each object is as a node.With among the object order result in center object and the list object former as object to be showed.
Step S430 plays up according to the object's position to be showed of setpoint distance after to initialization.The distance of calculated relationship two end nodes interval and any two nodes if spacing distance is not the distance of setting, is provided with an acceleration that moves to setpoint distance for two nodes, according to the acceleration of each node, calculates the speed of each node.Mobile node, if node has moved to the border of demonstration, the speed of inversion node moves in the border.Along with the complexity increase of object relationship, adopt the object relationship layout of optimizing, to experience better, information is more clear, accurate.
In one embodiment, the method for above-mentioned displaying object relationship also be included in obtain wait to show object after, judge to wait to show whether object is the duplication of name object, if the duplication of name object is handled.As shown in Figure 4, specifically comprise:
Step S21 sets up duplication of name list object information.In the present embodiment, be example, go out differentiable information between well-known internet personage that industry-by-industry bears the same name and the duplication of name personage by manual sorting with personage.Take following format record: " personage's name # distinguishes level # classification # and distinguishes information 1# differentiation information 2# ... "As Sun Yue # physical culture # basketball; Sun Yue # amusement # singer.Article # amusement # singer # Hong Kong; Article # amusement # singer # continent.Wherein, distinguish level be meant two duplication of name personages thoroughly can be distinguished after, the level degree of depth of related classification, the i.e. fine granularity that can distinguish for the personage of same name.For example for " Sun Yue " such people, distinguishing level is 1, because the name philtrum has only two " Sun Yue ", one is sport category, and one is the amusement class.If but information " article # amusement # singer # Hong Kong is arranged; Article # amusement # singer # continent ", so for " article " such people, it distinguishes level then is 2; because these two all belonged to the amusement class together by to be made the people of " article ", can't segment this two people only according to the amusement class, need go deep into one deck again; analyzing one is Hong Kong, and one is the continent.Differentiation information is meant and the information that two duplication of name personages distinguish can be gathered, promptly help to distinguish personage's of the same name keyword, these information appear in the residing context of personage, help systematic analysis and draw the concrete affiliated classification of this personage, as " basketball " in the above-mentioned example and " singer ".
Step S22 makes up the classification information hierarchical tree according to the internet historical information.As shown in Figure 5, specifically comprise the steps:
Step S220 obtains the internet historical information and extract classification information from described internet historical information, and classification information is carried out pre-service, obtains the classification information set of strings.Wherein, classification information in the internet information is when grasping internet information by search engine, according to the entry address of internet and the navigation information on the page, grasp, analyze the affiliated classification information of the internet information that obtains, classification information comprises class label and special list separator composition, generally, the class label behind the special separator is the subclass of the class label before the separator.Obtain the bit string S set of consolidation form by the classification information of pre-service internet information.Be class label as C, D, P, the consolidation form of formation is " C:P, D:P ", and then C is the parent class of D, and D is the subclass of C.
Step S221 resolves classification information string in the described classification information set of strings, calculates the probability of any classification set membership in the described classification information string.Detailed process is: at first resolve the classification information string in the above-mentioned classification information set of strings, add up the number of times of parent with respect to subclass, and draw the probability of subclass with respect to parent class.For any classification C and D, (C, D), expression C is as the number of times of the parent class of D to count T.
By probability normalization, obtain P (C, D), promptly classification C is that the probability of subclass of classification D is as follows:
Wherein, P (C, D) expression classification C is as the probability of the subclass of classification D, T (C, D) the classification C that obtains for aforementioned calculation is as the number of times of the subclass of classification D,
Expression classification D is as the number of times sum of the subclass of other classification.
Step S222 according to the probability initialization classification information hierarchical tree of described classification information string and any classification set membership, chooses the set of root classification.When classification E is the root classification, then do not have any classification F (F is not equal to E), make P (F, E)＞0.
Step S223, the arbitrary classification of recurrence expands to the path of root classification, obtain in such path that is clipped to root classification process all probability values of taking advantage of of two-layer classification set membership up and down, the probability that calculates this classification and be the subclass of root classification is clipped to the summation of probability value of taking advantage of described in all paths of root classification process for such.Specifically comprise: expansion particular category N, the possible parent class of institute of extraction classification N obtains gathering G; N is to the path of gathering each the element g among the G for the expansion classification, upgrading probability is P=P*P (g, N), recurrence expansion g is up to arriving certain root classification or can not expanding again, calculate classification N to element g, arrive in the path of root classification process probability value of taking advantage of of two-layer classification set membership about all again by element g; Finish recursive procedure, arrive each root classification, the probable value of this root classification of arrival on all paths of adding up is the probability that this subclass is the subclass of this root classification.Root classification set in this classification level of information tree comprises classification A and classification G, for any classification wherein, classification F for example, the process of probability of calculating classification F and be the subclass of classification A is: recurrence classification F is to the path of root classification A, comprising F-A and F-C-A two paths, probability value of taking advantage of of two-layer classification set membership about obtaining in this two paths, wherein this probability value of taking advantage of is 0.4 in the F-A path, and this probability value of taking advantage of is 0.3*0.5=0.15 in the F-C-A path, and then calculating classification F is the summation (being 0.4+0.15=0.55) of above-mentioned two probability values of taking advantage of that obtain as the probability of the subclass of root classification A; Recurrence classification F comprises the F-H-G path to the path of root classification G, and the probability value of taking advantage of in this path is 0.5*0.4=0.2, and then classification F is that the probability of the subclass of root classification G is 0.2.The probability of the subclass as the root classification of all categories can be used in the process of information category under the pending duplication of name object of follow-up judgement in the classification information hierarchical tree that calculates.
Step S23 handles the duplication of name object according to the classification information hierarchical tree that makes up.As shown in Figure 6, concrete steps comprise:
Step S230 judges describedly wait to show whether object is present in the duplication of name list object information, if, execution in step S231, otherwise finish.In the present embodiment, be object, judge whether internet personage is present in the duplication of name personage tabulation with personage.
Step S231 extracts the classification information string of waiting to show object place internet information.
Step S232 judges whether the classification information string exists, if, execution in step S233 then, if not, execution in step S234 then.
Step S233 according to described classification information hierarchical tree, obtains the probability of the subclass as the root classification of all categories in the described classification information string, and the root classification of getting described probability maximum is as the output result.According to the classification information hierarchical tree, calculate in the classification information string each classification as the probability of the subclass of root classification.
Step S234, the information distinguished of duplication of name object in the retrieval internet information, and judge whether and can distinguish, if, execution in step S235, otherwise finish.
Step S235, the object to be showed that output is corresponding.
In one embodiment, as shown in Figure 7, a kind of system that shows object relationship comprises information acquisition module 10, processing module 20, conjunctive word acquisition module 30 and output display module 40.Wherein,
Be equivalent to, occurred the day before yesterday 4 times, be the contribution of last weight
In like manner, 5 contribution margins to last comprehensive weight that occur with today for 3 times of appearance yesterday are respectively 0.75 and 5.0.
By calculating the close degree of personage's relation, the employing weighted value is represented, by comparing the size of weighted value, weighted value is sorted again, obtain the bigger personage of weighted value, personage's (object promptly to be showed) of the bigger weighted value that obtains is sent to front-end user interface.In addition, also regularly upgrade the personage concerns circle to processing module 20, by the default time, and timing acquiring people information, the personage that upgrades in time relation.Time can be set at 5 minutes upgrade once or upgraded in 1 hour one inferior.
Conjunctive word acquisition module 30 is used for according to waiting to show that object obtains treating in target date from internet information and shows the object relationship conjunctive word.Conjunctive word acquisition module 30 is by summing up the object relationship conjunctive word of the Rule Extraction sentence of associated extraction to Chinese sentence part of speech combination rule, characteristic of speech sounds in the internet information.In one embodiment, conjunctive word acquisition module 30 is further used for choosing the candidate association speech of waiting to show object relationship, calculates the candidate association speech weight wait to show object relationship and conjunctive word timeliness weight and draws and wait to show the object relationship conjunctive word in target date.The formula of the candidate association speech weight of calculating object relation is as follows:
F(w，d)＝log(length(w))*P(TF(w)*IDF(w))*H(w)*Title(w)
Wherein, (d represents the date to F for w, the d) weighted value of expression conjunctive word w; Length (w) is the length of conjunctive word, has adopted the length priority principle in the present embodiment, and based on a basic hypothesis: the more semanteme of horn of plenty generally can be expressed in the Chinese phrase that number of words is many more.TF (w) is the word frequency of conjunctive word w in internet information, and IDF (w) is the information that obtains of adding up in advance, is meant the contrary document frequency of speech w.Function P is a piecewise function.H (w) is regular weight, the extraction of conjunctive word mainly depends on the Chinese sentence syntax rule, and the word that each Rule Extraction is come out can be endowed an initial weighted value, is regular weight, conjunctive word w is extracted by Different Rule and obtains, and its regular weight H value is also different.Title (w) is meant the weight that conjunctive word w increases when appearing at the title of internet information, and default value is 1.Conjunctive word that calculates and weighted value thereof deposit in the Database Systems, as the candidate word of conjunctive word between two objects that extract in the sentence.
In addition, conjunctive word acquisition module 30 is by introducing time attenuation function, and calculated candidate conjunctive word timeliness weight is equivalent to the conjunctive word weight and multiply by a variable time factor.Formula is as follows:
Wherein, α and β are adjustable parameter, Dis tan ce (CurrentDay, d) time gap of expression date d and current date.In the present embodiment, closing with the personage is example, for two personages, can utilize following formula to calculate the timeliness weight S (w) of each conjunctive word w between two personages:
Get access to the timeliness weight of all candidate association speech between two personages, candidate association speech timeliness weight is sorted from big to small, can select several of coming the front and output to the user, as describing the conjunctive word that concerns between two personages.
In one embodiment, as shown in Figure 8, output display module 40 comprises that also input block 400, list object set up unit 401, wait to show object's position initialization unit 402 and object's position rendering unit 403.Wherein, input block 400 is used to receive user's inputted search object.List object is set up unit 401 and is used for setting up list object and obtaining object to be showed and the object relationship conjunctive word according to object search.Wait to show that object's position initialization unit 402 is used for showing according to waiting of obtaining that object and object relationship conjunctive word are treated and shows that object's position carries out the initialization setting.Object's position rendering unit 403 is used for playing up according to the object's position to be showed of setpoint distance after to initialization.The detailed process of playing up comprises: the distance of calculated relationship two end nodes interval and any two nodes, if spacing distance is not the distance of setting, an acceleration that moves to setpoint distance is set for two nodes,, calculates the speed of each node according to the acceleration of each node.Mobile node, if node has moved to the border of demonstration, the speed of inversion node moves in the border.
In one embodiment, as shown in Figure 9, the system of above-mentioned displaying object relationship also comprises duplication of name object handles module 50.Duplication of name object handles module 50 is used for after obtaining object, judges that whether object is the duplication of name object and the duplication of name object handled.
In one embodiment, duplication of name object handles module 50 is further used for setting up duplication of name list object information, makes up the classification information hierarchical tree according to historical internet historical information, according to the classification information hierarchical tree that makes up the duplication of name object is handled.
In another embodiment, as shown in figure 10, duplication of name object handles module 50 makes up classification information hierarchical tree part, specifically comprises:
Classification information acquiring unit 500 extracts classification information from the internet historical information.
Classification information string in the classification information set of strings is resolved in probability calculation unit 502, calculates the probability of any classification set membership in the described classification information string.Detailed process is: the classification information string in the above-mentioned classification information set of strings is at first resolved in probability calculation unit 502, adds up the number of times of parent with respect to subclass, and draws the probability of subclass with respect to parent class.For any classification C and D, (C, D), expression C is as the number of times of the parent class of D to count T.By probability normalization, obtain P (C, D), promptly classification C is that the probability of subclass of classification D is as follows:
Wherein, P (C, D) expression classification C is as the probability of the subclass of classification D, T (C, D) the classification C that obtains for aforementioned calculation is as the number of times of the subclass of classification D,
Expression classification D is as the number of times sum of the subclass of other classification.
Classification information hierarchical tree initialization unit 503 according to the probability initialization classification information hierarchical tree of described classification information string and any classification set membership, is chosen the set of root classification.When classification E is the root classification, then do not have any classification F (F is not equal to E), make P (F, E)＞0.
In another embodiment, duplication of name object handles module 50 is further used for judging waits in the internet information to show that object is in duplication of name list object information, extract the classification information string of object place internet information, calculate the probability of the subclass as the root classification of all categories in the described classification information, the root classification of getting described probability maximum is as the output result.
The method and system of above-mentioned displaying object relationship, adopt the object relationship of calculating in target date to spend closely and draw object to be showed, and obtain the object relationship conjunctive word in target date, can upgrade timely and wait to show object and wait to show the object relationship conjunctive word, can show relation between objects more timely like this, embody the ageing of object relationship.
In addition, the duplication of name object is distinguished judgement, improved the accuracy of waiting to show object and object relationship, allow the user better be experienced; Employing is played up the position of object search, and the flexible variation of holding position can embody object relationship clear, accurately.
The above embodiment has only expressed several embodiment of the present invention, and it describes comparatively concrete and detailed, but can not therefore be interpreted as the restriction to claim of the present invention.Should be pointed out that for the person of ordinary skill of the art without departing from the inventive concept of the premise, can also make some distortion and improvement, these all belong to protection scope of the present invention.Therefore, the protection domain of patent of the present invention should be as the criterion with claims.
Claims (18)
1. method of showing object relationship may further comprise the steps:
A. from internet information, obtain the object and the acquisition target co-occurrence frequency;
B. calculate total frequency that target date, object occurred in twos according to the described co-occurrence frequency, again according to total frequency calculating object degree in close relations and draw object to be showed;
C. according to waiting to show that object obtains treating in target date and shows the object relationship conjunctive word from internet information;
D. wait to show object and wait to show that the object relationship conjunctive word exports displaying according to described.
2. the method for displaying object relationship according to claim 1 is characterized in that, the computing formula that object relationship is spent closely among the described step B is:
Wherein, R (A B) is the degree in close relations of any object A, B, and F (A, B, D) the total frequency that occurs in twos for object A, the B of date D, α and β are adjustable parameter, Dis tan ce (CurrentDay, D) time gap of expression date D and current date.
3. the method for displaying object relationship according to claim 1 is characterized in that, step C comprises:
C1. choose the candidate association speech of waiting to show object relationship;
C2. calculate the weight of the candidate association speech wait to show object relationship;
C3. the timeliness weight of calculating the candidate association speech wait to show object relationship draws waits to show the object relationship conjunctive word in target date.
4. the method for displaying object relationship according to claim 3 is characterized in that, the formula that calculates the weight of the candidate association speech wait to show object relationship among the step C2 is:
F(w，d)＝log(length(w))*P(TF(w)*IDF(w))*H(w)*Title(w)
Wherein, (d represents the date to F for w, the d) weighted value of expression conjunctive word w; Length (w) is the length of conjunctive word, and TF (w) is the word frequency of conjunctive word w in internet information, and IDF (w) is meant the contrary document frequency of speech w; Function P is a piecewise function; H (w) is regular weight; Title (w) is meant the weight that conjunctive word w increases when appearing at the title of internet information, and default value is 1.
5. the method for displaying object relationship according to claim 1, it is characterized in that the method for described displaying object relationship also comprises step e: obtain wait to show object after, judge and describedly wait to show whether object is the duplication of name object, if the duplication of name object is handled.
6. the method for displaying object relationship according to claim 5 is characterized in that, step e further may further comprise the steps:
E1. set up duplication of name list object information;
E2. make up the classification information hierarchical tree according to the internet historical information;
E3. according to the classification information hierarchical tree that makes up the duplication of name object is handled.
7. the method for displaying object relationship according to claim 6 is characterized in that, step e 2 comprises:
E21. obtain the internet historical information and from described internet historical information, extract classification information, and classification information is carried out pre-service, obtain the classification information set of strings;
E22. resolve classification information string in the described classification information set of strings, calculate the probability of any classification set membership in the described classification information string;
E23. according to the probability initialization classification information hierarchical tree of described classification information string and any classification set membership, choose the set of root classification;
E24. the arbitrary classification of recurrence expands to the path of root classification, obtain in such path that is clipped to root classification process all probability values of taking advantage of of two-layer classification set membership up and down, the probability that calculates this classification and be the subclass of root classification is clipped to the summation of probability value of taking advantage of described in all paths of root classification process for such.
8. the method for displaying object relationship according to claim 7 is characterized in that, step e 3 comprises:
E30. judge describedly wait to show whether object is present in the duplication of name list object information, if, execution in step E31, otherwise finish;
E31. extract the classification information string of waiting to show object place internet information;
E32. judge whether described classification information string exists, if, execution in step E33 then, if not, execution in step E34 then;
E33. according to described classification information hierarchical tree, obtain the probability of the subclass as the root classification of all categories in the described classification information string, the root classification of getting described probability maximum is as the output result;
E34. retrieve the information distinguished of duplication of name object in the internet information, and judge whether and can distinguish, if, execution in step E35, otherwise finish;
E35. export corresponding object to be showed.
9. more ask the method for 1 described displaying object relationship according to right, it is characterized in that step D specifically comprises:
D1. receive user's inputted search object;
D2. set up list object according to object search and obtain object to be showed and the object relationship conjunctive word;
D3. show according to waiting of obtaining that object and object relationship conjunctive word are treated and show that object's position carries out the initialization setting;
D4. play up according to the object's position to be showed of setpoint distance after initialization.
10. a system that shows object relationship is characterized in that, this system comprises:
Information acquisition module is used for obtaining the object and the acquisition target co-occurrence frequency from internet information;
Processing module is calculated total frequency that target date, object occurred in twos, again according to total frequency calculating object degree in close relations, and draws object to be showed;
The conjunctive word acquisition module is used for according to waiting to show that object obtains treating in target date from internet information and shows the object relationship conjunctive word;
The output display module is according to waiting to show object and waiting to show that the object relationship conjunctive word exports displaying.
11. the system of displaying object relationship according to claim 10 is characterized in that, the formula that described processing module is calculated the close degree of described object relationship is:
Wherein, (A B) is the degree weight in close relations of object A, B to R, and (A, B D) are the object A of date D, sum frequency that B occurs in twos to F, and α and β are adjustable parameter, Dis tan ce (CurrentDay, D) time gap of expression date D and current date.
12. the system of displaying object relationship according to claim 10, it is characterized in that, the conjunctive word acquisition module is further used for choosing the candidate association speech of waiting to show object relationship, calculates the candidate association speech weight wait to show object relationship and conjunctive word timeliness weight and draws and wait to show the object relationship conjunctive word in target date.
13. the system of displaying object relationship according to claim 12 is characterized in that, the weight formula of the candidate association speech of described conjunctive word acquisition module calculating object relation is:
F(w，d)＝log(length(w))*P(TF(w)*IDF(w))*H(w)*Title(w)
Wherein, (d represents the date to F for w, the d) weighted value of expression conjunctive word w; Length (w) is the length of conjunctive word, and TF (w) is the word frequency of conjunctive word w in internet information, and IDF (w) is meant the contrary document frequency of speech w; Function P is a piecewise function; H (w) is regular weight; Title (w) is meant the weight that conjunctive word w increases when appearing at the title of internet information, and default value is 1.
14. the system of displaying object relationship according to claim 10, it is characterized in that, described system also comprises duplication of name object handles module, and described duplication of name object handles module is used for after obtaining object, judges that whether object is the duplication of name object and the duplication of name object handled.
15. the system of displaying object relationship according to claim 14, it is characterized in that, described duplication of name object handles module is further used for setting up duplication of name list object information, make up the classification information hierarchical tree according to historical internet historical information, the duplication of name object is handled according to the classification information hierarchical tree that makes up.
16. the system of displaying object relationship according to claim 15 is characterized in that, described duplication of name object handles module further comprises:
The classification information acquiring unit extracts classification information from the internet historical information;
Pretreatment unit carries out pre-service to described classification information, obtains the classification information set of strings;
Classification information string in the classification information set of strings is resolved in the probability calculation unit, calculates the probability of any classification set membership in the described classification information string;
Classification information hierarchical tree initialization unit according to the probability initialization classification information hierarchical tree of described classification information string and any classification set membership, is chosen the set of root classification;
The recurrence expanding element, be used for the path that the arbitrary classification of recurrence expands to the root classification, obtain in such path that is clipped to root classification process all probability values of taking advantage of of two-layer classification set membership up and down, the probability that calculates this classification and be the subclass of root classification is clipped to the summation of probability value of taking advantage of described in all paths of root classification process for such.
17. the system of displaying object relationship according to claim 16, it is characterized in that, described duplication of name object handles module is further used for judging waits in the internet information to show that object is in duplication of name list object information, extract the classification information string of object place internet information, calculate the probability of the subclass as the root classification of all categories in the described classification information, the root classification of getting described probability maximum is as the output result.
18. the system of displaying object relationship according to claim 10 is characterized in that, described output display module also comprises:
Input block is used to receive user's inputted search object;
The unit is set up in tabulation, is used for setting up list object and obtaining object to be showed and the object relationship conjunctive word according to object search;
Wait to show the object's position initialization unit, be used for showing according to waiting of obtaining that object and object relationship conjunctive word are treated and show that object's position carries out the initialization setting;
The object's position rendering unit is used for playing up according to the object's position to be showed of setpoint distance after to initialization.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 201010146061 CN102214186B (en) | 2010-04-07 | 2010-04-07 | Method and system for displaying object relation |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 201010146061 CN102214186B (en) | 2010-04-07 | 2010-04-07 | Method and system for displaying object relation |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN102214186A true CN102214186A (en) | 2011-10-12 |
| CN102214186B CN102214186B (en) | 2013-04-24 |
Family
ID=44745501
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN 201010146061 Active CN102214186B (en) | 2010-04-07 | 2010-04-07 | Method and system for displaying object relation |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102214186B (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104899241A (en) * | 2014-03-07 | 2015-09-09 | 歌乐株式会社 | Related data generating apparatus and related data generating method |
| CN106095748A (en) * | 2016-06-06 | 2016-11-09 | 东软集团股份有限公司 | A kind of method and device generating event relation collection of illustrative plates |
| CN106101748A (en) * | 2016-07-20 | 2016-11-09 | 东软集团股份有限公司 | Program processing method and device |
| CN106407210A (en) * | 2015-07-29 | 2017-02-15 | 阿里巴巴集团控股有限公司 | Display method and device of business object |
| CN108830735A (en) * | 2017-12-27 | 2018-11-16 | 广州市云润大数据服务有限公司 | Interpersonal relationships analysis method and system on a kind of line |
| CN109766786A (en) * | 2018-12-21 | 2019-05-17 | 深圳云天励飞技术有限公司 | Character relation analysis method and Related product |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2002288166A (en) * | 2001-03-26 | 2002-10-04 | Ricoh Co Ltd | Relevant information extraction method and semantic information extraction method |
| CN101436186A (en) * | 2007-11-12 | 2009-05-20 | 北京搜狗科技发展有限公司 | Method and system for providing related searches |
| CN101520785A (en) * | 2008-02-29 | 2009-09-02 | 富士通株式会社 | Information retrieval method and system therefor |
-
2010
- 2010-04-07 CN CN 201010146061 patent/CN102214186B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2002288166A (en) * | 2001-03-26 | 2002-10-04 | Ricoh Co Ltd | Relevant information extraction method and semantic information extraction method |
| CN101436186A (en) * | 2007-11-12 | 2009-05-20 | 北京搜狗科技发展有限公司 | Method and system for providing related searches |
| CN101520785A (en) * | 2008-02-29 | 2009-09-02 | 富士通株式会社 | Information retrieval method and system therefor |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104899241A (en) * | 2014-03-07 | 2015-09-09 | 歌乐株式会社 | Related data generating apparatus and related data generating method |
| US10157348B2 (en) | 2014-03-07 | 2018-12-18 | Clarion Co., Ltd. | Related data generating apparatus, related data generating method, and program |
| CN106407210A (en) * | 2015-07-29 | 2017-02-15 | 阿里巴巴集团控股有限公司 | Display method and device of business object |
| CN106095748A (en) * | 2016-06-06 | 2016-11-09 | 东软集团股份有限公司 | A kind of method and device generating event relation collection of illustrative plates |
| CN106095748B (en) * | 2016-06-06 | 2019-08-27 | 东软集团股份有限公司 | A kind of method and device generating event relation map |
| CN106101748A (en) * | 2016-07-20 | 2016-11-09 | 东软集团股份有限公司 | Program processing method and device |
| CN106101748B (en) * | 2016-07-20 | 2020-04-28 | 东软集团股份有限公司 | Program processing method and device |
| CN108830735A (en) * | 2017-12-27 | 2018-11-16 | 广州市云润大数据服务有限公司 | Interpersonal relationships analysis method and system on a kind of line |
| CN109766786A (en) * | 2018-12-21 | 2019-05-17 | 深圳云天励飞技术有限公司 | Character relation analysis method and Related product |
| CN109766786B (en) * | 2018-12-21 | 2020-10-23 | 深圳云天励飞技术有限公司 | Character relation analysis method and related product |
Also Published As
| Publication number | Publication date |
|---|---|
| CN102214186B (en) | 2013-04-24 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106709040B (en) | Application search method and server | |
| CN106649818B (en) | Application search intention identification method and device, application search method and server | |
| CN103514299B (en) | Information search method and device | |
| CN106970991B (en) | Similar application identification method and device, application search recommendation method and server | |
| CN103678576B (en) | The text retrieval system analyzed based on dynamic semantics | |
| US8645385B2 (en) | System and method for automating categorization and aggregation of content from network sites | |
| CN106682169B (en) | Application label mining method and device, application searching method and server | |
| CN102929873B (en) | Method and device for extracting searching value terms based on context search | |
| CN104199833B (en) | The clustering method and clustering apparatus of a kind of network search words | |
| US20140201180A1 (en) | Intelligent Supplemental Search Engine Optimization | |
| CN106339502A (en) | Modeling recommendation method based on user behavior data fragmentation cluster | |
| CN102915380A (en) | Method and system for carrying out searching on data | |
| WO2011080899A1 (en) | Information recommendation method | |
| CN106663117A (en) | Constructing a graph that facilitates provision of exploratory suggestions | |
| CN106204156A (en) | A kind of advertisement placement method for network forum and device | |
| CN102214186B (en) | Method and system for displaying object relation | |
| CN104102721A (en) | Method and device for recommending information | |
| CN102930054A (en) | Data search method and data search system | |
| CN102163198A (en) | A method and a system for providing new or popular terms | |
| CN107799116A (en) | More wheel interacting parallel semantic understanding method and apparatus | |
| CN107748745B (en) | Enterprise name keyword extraction method | |
| CN112989208B (en) | Information recommendation method and device, electronic equipment and storage medium | |
| CN101923556B (en) | Method and device for searching webpages according to sentence serial numbers | |
| CN107688616A (en) | Show unique fact of entity | |
| KR20080037413A (en) | On line context aware advertising apparatus and method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant |