CN106446157A - Route destination recommending method and device - Google Patents
Route destination recommending method and device Download PDFInfo
- Publication number
- CN106446157A CN106446157A CN201610844639.7A CN201610844639A CN106446157A CN 106446157 A CN106446157 A CN 106446157A CN 201610844639 A CN201610844639 A CN 201610844639A CN 106446157 A CN106446157 A CN 106446157A
- Authority
- CN
- China
- Prior art keywords
- stroke destination
- candidate
- weight
- stroke
- destination
- 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
- 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
Abstract
The invention discloses a route destination recommending method and device. The route destination recommending method comprises the following steps: confirming a regular route destination of a user according to user track data; confirming a recent preferred route destination of the user according to user retrieval data; confirming candidate route destinations according to the regular route destination and the recent preferred route destination; and respectively confirming a confidence coefficient score of each candidate route destination, ordering the candidate route destinations in a sequence of high scores to small scores, and recommending candidate route destinations of the front M positions after ordering to the users, wherein M is a positive integer. By adopting the scheme disclosed by the invention, the accuracy of recommended results and the like can be improved.
Description
【Technical field】
The present invention relates to Internet technology, more particularly to stroke destination recommend method and apparatus.
【Background technology】
In order to intellectuality, user is helped to secretaryization to complete process of entirely going on a journey, lifting user uses map products
Viscosity and satisfaction, it is desirable to be able to intelligently find the possible stroke destination of user and recommended.
In prior art, it is typically only capable to for the family of the user for excavating and company to recommend user as stroke destination,
Cannot be recommended according to the actual demand of user, so as to reduce the accuracy of recommendation results.
【Content of the invention】
The invention provides method and apparatus is recommended by stroke destination, it is possible to increase the accuracy of recommendation results.
Concrete technical scheme is as follows:
Method is recommended by a kind of stroke destination, including:
The regular stroke destination of user is determined according to user trajectory data, and is determined according to user search data
The recent interest stroke destination of the user;
Candidate's stroke destination is determined according to the regularity stroke destination and the recent interest stroke destination;
The confidence score of each candidate stroke destination is determined respectively, and according to scoring order from high to low to each
Candidate's stroke destination is ranked up, and the candidate's stroke destination after sequence in front M position is recommended the user, and M is for just
Integer.
A kind of stroke destination recommendation apparatus, including:First processing units, second processing unit, the 3rd processing unit with
And recommendation unit;
The first processing units, for determining the regular stroke destination of user according to user trajectory data, and
It is sent to the 3rd processing unit;
The second processing unit, for determining the recent interest stroke purpose of the user according to user search data
Ground, and it is sent to the 3rd processing unit;
3rd processing unit, for according to the regularity stroke destination and the recent interest stroke destination
Determine candidate's stroke destination;
The recommendation unit, for determining the confidence score of each candidate's stroke destination respectively, and according to scoring
Order from high to low is ranked up to each candidate's stroke destination, and the candidate's stroke destination after sequence in front M position is pushed away
Recommend to the user, M is positive integer.
Can be seen that using scheme of the present invention by above-mentioned introduction, use can be determined according to user trajectory data
The regular stroke destination at family, determines the recent interest stroke destination of user, and then basis according to user search data
Regular stroke destination and recent interest stroke destination determine candidate's stroke destination, so, when needs are pushed away to user
When stroke destination is recommended, the confidence score of each candidate stroke destination can be determined respectively, and according to scoring from high to low
Order each candidate's stroke destination is ranked up, and then the candidate's stroke destination after sequence in front M position is recommended
User, compared to the mode for only recommending family and company in prior art, can be according to the reality of user in mode of the present invention
Demand etc. being recommended, so as to improve accuracy of recommendation results etc..
【Description of the drawings】
Fig. 1 is the flow chart that embodiment of the method is recommended by stroke destination of the present invention.
Fig. 2 is that the method for the regular stroke destination for determining user according to user trajectory data of the present invention is implemented
The flow chart of example.
Fig. 3 is the method reality of the recent interest stroke destination for determining user according to user search data of the present invention
The flow chart for applying example.
Fig. 4 is the strong time attribute weight for determining candidate's stroke destination and candidate's stroke destination of the present invention
Process schematic.
Fig. 5 is the process schematic to user's recommended candidate stroke destination of the present invention.
Fig. 6 is the composition structural representation of stroke destination recommendation apparatus embodiment of the present invention.
【Specific embodiment】
In order that technical scheme is clearer, clear, develop simultaneously embodiment referring to the drawings, to institute of the present invention
The scheme of stating is described in further detail.
Embodiment one
Fig. 1 is the flow chart that embodiment of the method is recommended by stroke destination of the present invention, as shown in figure 1, including following tool
Body implementation:
The regular stroke destination of user in 11, is determined according to user trajectory data;
The recent interest stroke destination of user in 12, is determined according to user search data;
In 13, candidate's stroke destination is determined according to regular stroke destination and recent interest stroke destination;
In 14, the confidence score of each candidate stroke destination determined respectively, and according to scoring from high to low
Order is ranked up to each candidate's stroke destination, and the candidate's stroke destination after sequence in front M position is recommended user, M
For positive integer.
Individually below above-mentioned implementing for each step is described in detail.
1) the regular stroke destination of user is determined according to user trajectory data
For regular stroke destination is obtained, user trajectory data can be obtained first, afterwards by user trajectory data
Be analyzed, determine the dwell point in user trajectory, so filter out temporal regularity from dwell point often go to place, make
Regular stroke destination for user.
Fig. 2 is that the method for the regular stroke destination for determining user according to user trajectory data of the present invention is implemented
The flow chart of example, as shown in Fig. 2 including implementation in detail below.
In 21, user trajectory data are obtained.
The user trajectory data for getting may include:Use data of the user to map such as Baidu map, and, by means of
User trajectory data that such as Baidu's software development kit (SDK) is collected by other application etc..
User trajectory data how to be obtained for prior art, the passing all of user trajectory data of user can be obtained.
In 22, by being analyzed to user trajectory data, the dwell point in user trajectory is determined.
How user trajectory data are analyzed being similarly prior art with the dwell point that determines in user trajectory, than
Such as, networking type can be positioned according to user, whether have the information such as base station switching to excavate the dwell point in user trajectory, including
The position coordinateses of dwell point, user are in the time of occurrence of dwell point, stay time etc..
In 23, the dwell point of noise types is filtered out.
Whether the operation of the dwell point that execution filters out noise types is optional.
Filtering rule can be pre-set, to filter to the dwell point of noise types.
Specifically include in filtering rule which content can be decided according to the actual requirements, such as, can be according to user in dwell point
Stay time or user in the time of occurrence etc. of dwell point, filter out the dwell point of noise types.
Such as, user has only stopped three minutes in a certain dwell point, then is likely to be user and gets on roadside on next road
Convenience store bought bottle beverage, for this kind of dwell point, then it is believed which is the dwell point of noise types.
In 24, dwell point close for geographical position is clustered.
After the process in 23, remaining be valuable dwell point, afterwards, can be sat according to the position of each dwell point
Mark, using existing clustering algorithm, dwell point close for geographical position is clustered.
Optionally, for arbitrary cluster result, if including stop points very few, may filter that cluster knot
Really, i.e., subsequent treatment is not carried out to the cluster result.
In 25, for each cluster result, 26~28 are executed respectively.
In 26, using the average of the position coordinateses of each dwell point in the cluster result as the corresponding row of the cluster result
The position coordinateses of journey destination.
The position coordinateses of each dwell point in the cluster result are known, then can calculate the equal of each position coordinate
Value, using result of calculation as the corresponding stroke destination of the cluster result position coordinateses.
The Annual distribution of each dwell point in the cluster result in 27, is occurred according to user, determines the cluster knot
The day rank regularity weight and all rank regularity weights of really corresponding stroke destination.
How to determine that day rank regularity weight and all rank regularity weights can be decided according to the actual requirements, be not limited to
In the following ways.
For day rank regularity weight, can be occurred according to user first each dwell point in the cluster result when
Between, the natural law that nearest scheduled duration such as nearest 14 days (two weeks) interior user reached the dwell point in the cluster result is counted,
Then with the natural law for counting divided by 14, so as to obtain the day rank regularity power of the corresponding stroke destination of the cluster result
Weight.
Assume the dwell point that user has reached in the cluster result in nearest 14 days for 7 days, then the day level for obtaining
Regular weight is not then 0.5, it can be seen that in the manner described above, and the value of the day rank regularity weight for obtaining is minimum
0, it is 1 to the maximum.
For all ranks regularity weight, can be occurred according to user respectively each dwell point in the cluster result when
Between, the Monday user for counting nearest scheduled duration such as nearest 4 weeks reached the natural law of the dwell point in the cluster result, and
With statistical result divided by 4, so as to all ranks regularity weight of Monday is obtained, similarly, Tuesday can be respectively obtained to star
All ranks regularity weight of day phase, then selects a maximum in all ranks regularity weight from Monday to Sunday
Value, using the maximum as the regular weight of all ranks of the corresponding stroke destination of the cluster result.
Assume, in 4 Mondays in nearest 4 weeks, have 2 Monday users that the dwell point in the cluster result was reached,
All ranks regularity weight of the Monday for so obtaining is then that all ranks for obtaining are advised 0.5, similarly, in the manner described above
The value minimum 0 of rule property weight, is 1 to the maximum.
Monday is being obtained to after 7 all rank regularity weights of Sunday, maximum therein can gathered as this
All ranks regularity weight of the corresponding stroke destination of class result.
In 28, rank regularity weight in day is compared with corresponding first threshold, by all ranks regularity weight
It is compared with corresponding Second Threshold, if arbitrary weight is more than corresponding threshold value, by corresponding for cluster result stroke mesh
Ground be defined as regular stroke destination, and will be greater than the rule of the weight as the regular stroke destination of corresponding threshold value
Rule property weight, abandons the weight less than or equal to corresponding threshold value.
In the day rank regularity weight for obtaining the corresponding stroke destination of the cluster result and all rank regularity weights
Afterwards, respectively two weights can be compared with corresponding threshold value, will day rank regularity weight carry out with first threshold
Relatively, all ranks regularity weight is compared with Second Threshold, if rank regularity weight in day is more than first threshold, or
Person, all rank regularity weights are more than Second Threshold, or, rank regularity weight in day is more than first threshold and all rank rules
Property weight be more than Second Threshold, then corresponding for cluster result stroke destination is defined as regular stroke destination.
For the regular stroke destination, may only include in its regular weight a day rank regularity weight,
It is likely to only include all ranks regularity weight, it is also possible to while including day rank regularity weight and the regular power of all ranks
Weight.
Wherein, if comparative result be day rank regularity weight more than first threshold but all rank regularity weights less than or
Equal to Second Threshold, then a day rank regularity weight will be only included in the regular weight of the regular stroke destination;If ratio
Relatively result is day rank regularity weight less than or equal to first threshold but all rank regularity weights are more than Second Threshold, then should
All ranks regularity weight will be only included in the regular weight of regular stroke destination;If comparative result is day rank rule
Property weight is more than first threshold and all ranks regularity weight is more than Second Threshold, then the regularity of the regular stroke destination
Day rank regularity weight and all rank regularity weights will be included in weight simultaneously.
The concrete value of first threshold and Second Threshold can all be decided according to the actual requirements.
2) the recent interest stroke destination of user is determined according to user search data
For recent interest stroke destination is obtained, user search data, the user described in the present embodiment can be obtained first
Retrieval data refer in particular to retrieval data of the user in map, can determine the recent interest of user afterwards according to user search data
Stroke destination.
Fig. 3 is the method reality of the recent interest stroke destination for determining user according to user search data of the present invention
The flow chart for applying example, as shown in figure 3, including implementation in detail below.
In 31, user search data are obtained.
Such as, user can carry out route planning using map or place is checked, so as to carry out the inspection of point of interest (poi)
Rope, such as Beijing South Station, these are retrieval data.
User search data how to be obtained for prior art, the passing all of user search data of user can be obtained.
In 32, the retrieval data of noise types are filtered out.
Whether the operation of the retrieval data that execution filters out noise types is optional.
Filtering rule can be pre-set, to filter to the retrieval data of noise types, is specifically wrapped in filtering rule
Include which content can be decided according to the actual requirements.
In 33, the retrieval data of same for correspondence poi are clustered.
After the process in 32, remaining be valuable retrieval data, on this basis, can will correspondence same
The retrieval data of poi are clustered.
Such as, for this poi of Beijing South Station, user carried out repeatedly retrieval, then this repeatedly retrieves corresponding retrieval
Data will be clustered together.
Optionally, for arbitrary cluster result, if including retrieval data number very few, may filter that cluster knot
Really, i.e., subsequent treatment is not carried out to the cluster result.
In 34, for each cluster result, 35~36 are executed respectively.
In 35, according to retrieval type and the retrieval time of each retrieval data in the cluster result, the cluster is determined
As a result the recent interest weight of corresponding poi.
If it is determined that the recent interest weight of the corresponding poi of the cluster result can be decided according to the actual requirements, it is not limited to
In the following ways.
For each the retrieval data in the cluster result, can respectively according to its retrieve type and retrieval time with current
The distance between time, determine the scoring of retrieval type and the retrieval time scoring of the retrieval data.Retrieval type may include road
Line gauge is drawn and is checked with place, can set the corresponding retrieval type scoring of different retrieval types in advance respectively.Retrieval time away from
From current time more close to, the scoring of corresponding retrieval time is higher, conversely, retrieval time is more remote apart from current time, corresponding inspection
The scoring of rope time is lower, the corresponding relation between the distance between retrieval time and current time and the scoring of corresponding retrieval time
Can preset, such as, multiple continuous intervals can be marked off in advance, each interval corresponds to different retrievals respectively
Time scores, and the interval according to belonging to the distance between retrieval time and current time determines its corresponding retrieval time
Scoring.
After the retrieval type scoring for obtaining each retrieval data and retrieval time scoring, can be by the inspection of the retrieval data
Rope type scoring and retrieval time score respectively be added after corresponding multiplication, so as to obtain the scoring of the retrieval data,
The concrete value of each coefficient can be decided according to the actual requirements.
So, for the cluster result, can by including each retrieval data scoring be added, as the cluster
As a result the recent interest weight of corresponding poi.
In 36, recent interest weight is compared with corresponding 3rd threshold value, if interest weight is more than the 3rd in the recent period
Threshold value, then be defined as recent interest stroke destination by corresponding for cluster result poi.
After the recent interest weight for obtaining the corresponding poi of the cluster result, which can be compared with the 3rd threshold value,
If interest weight is more than the 3rd threshold value in the recent period, corresponding for cluster result poi is defined as recent interest stroke destination.
The concrete value of the 3rd threshold value can be decided according to the actual requirements.
3) candidate's stroke destination is determined according to regular stroke destination and recent interest stroke destination
According to 1) and 2) in mode determine respectively regular stroke destination and recent interest stroke destination it
Afterwards, candidate's stroke destination can be further determined that out, also, for realizing subsequent recommendation, in addition it is also necessary to determine each candidate respectively
The strong time attribute weight of stroke destination.
Fig. 4 is the strong time attribute weight for determining candidate's stroke destination and candidate's stroke destination of the present invention
Process schematic.
As shown in figure 4, each regular stroke destination is directed to, following process can be carried out respectively:
According to the position coordinateses of the regular stroke destination, determine on map and be located at the regular stroke destination week
Enclose the poi in the region of predefined size;
According to distance and poi temperature with the regular stroke destination, a representative is selected from the poi for determining
Property poi, using the representative poi as candidate's stroke destination, and using the regular weight of the regular stroke destination as
The regular weight of candidate's stroke destination.
Such as, can first determine that out on map centered on the position coordinateses of the regular stroke destination, with 100 meters and be
Poi in the border circular areas of radius, afterwards, can be according to the distance and poi temperature with the regular stroke destination, from determination
A representativeness poi is selected in the poi for going out, using the representative poi as candidate's stroke destination, i.e., one regular stroke mesh
Ground correspond to a representativeness poi, representativeness poi as candidate's stroke destination.
How to select representative poi to be decided according to the actual requirements, such as, have two poi selective, one is Bei Jingnan
Stand, another is that roadside convenience stores, the temperature of Beijing South Station is apparently higher than roadside convenience stores, then, even if roadside convenience stores phase
Than in Beijing South Station apart from the regular stroke destination closer to, can generally also select Beijing South Station as representative poi, because
This, can be the distance weight different with temperature setting, and comprehensive actual range, temperature and weight select representative poi.
As shown in figure 4, can be using 2) in all recent interest stroke destination that determines all as candidate's stroke destination,
And using the recent interest weight of recent interest stroke destination as candidate's stroke destination recent interest weight.
Afterwards, for each candidate's stroke destination, the description information of candidate's stroke destination can be obtained respectively, described
Description information may include type and label etc., then inquires about the strong time attribute weight table for pre-setting, finds out the candidate row
The corresponding strong time attribute weight of the description information of journey destination, using the strong time attribute weight for finding out as candidate's stroke
The strong time attribute weight of destination.
The strong time, according to different stroke destinatioies for the requirement stringency etc. of time, can be generated according to practical experience
Attribute weight table, the concrete form of form can be decided according to the actual requirements, but need to ensure for candidate's stroke destination,
One strong time attribute weight can only be found according to its description information, but different candidate's stroke destinatioies can correspond to identical
Strong time attribute weight.
4) to user's recommended candidate stroke destination
Fig. 5 is the process schematic to user's recommended candidate stroke destination of the present invention, as shown in figure 5, working as user
Spread out the map, when entering recommendation interface, you can carry out the recommendation of candidate's stroke destination for user.
Specifically, for each candidate's stroke destination, the weight information of candidate's stroke destination can be obtained first, is obtained
The weight that gets includes strong time attribute weight and one below or whole:Regular weight, recent interest weight, afterwards,
The confidence score of candidate's stroke destination can be determined according to all weights of candidate's stroke destination.
For each candidate's stroke destination, the weight for necessarily including is strong time attribute weight, and other weights
Then potentially include, it is also possible to do not include, such as interest stroke destination in the near future, recent interest weight is will also include, but
Regular weight will not generally be included, and for regular stroke destination, will also include regular weight, but generally not
Recent interest weight can be included.
A confidence calculations formula can be generated previously according to experience, so, for each candidate's stroke destination, you can
According to the confidence calculations formula and all weights of candidate's stroke destination, putting for candidate's stroke destination is calculated
Confidence score, the concrete form of the confidence calculations formula can be decided according to the actual requirements, such as, can respectively by each weight with
It is added after corresponding multiplication, if lacking a certain weight, it is believed that the weight is 0.
After the confidence score for respectively obtaining each candidate's stroke destination, order that can be according to scoring from high to low is right
Each candidate's stroke destination is ranked up, and the candidate's stroke destination after sequence in front M position is recommended user, and M is for just
Integer, concrete value can equally be decided according to the actual requirements, and such as value is 3.
It should be noted that as the user trajectory data of user and user search data are to constantly update, therefore,
1 can be periodically carried out)~3) etc. operation, such as in daily zero point, examined according to newest user trajectory data and user respectively
Rope data, re-execute 1)~3) etc. operation so that recommendation results are more and more accurate.
It is more than the introduction with regard to embodiment of the method, below by way of device embodiment, scheme of the present invention is entered to advance
One step explanation.
Embodiment two
Fig. 6 is the composition structural representation of stroke destination recommendation apparatus embodiment of the present invention, as shown in fig. 6, bag
Include:First processing units 61, second processing unit 62, the 3rd processing unit 63 and recommendation unit 64.
First processing units 61, for determining the regular stroke destination of user according to user trajectory data, concurrently
Give the 3rd processing unit 63;
Second processing unit 62, for determining the recent interest stroke destination of user according to user search data, and
It is sent to the 3rd processing unit 63;
3rd processing unit 63, for determining candidate according to regular stroke destination and recent interest stroke destination
Stroke destination;
Recommendation unit 64, for determining the confidence score of each candidate's stroke destination respectively, and according to scoring by
High to Low order is ranked up to each candidate's stroke destination, and the candidate's stroke destination after sequence in front M position is recommended
To user, M is positive integer.
First processing units 61 can obtain user trajectory data first, afterwards by being analyzed to user trajectory data,
Determine the dwell point in user trajectory, so filter out temporal regularity from dwell point often go to place, as user's
Regular stroke destination.
How user trajectory data are obtained for prior art, first processing units 61 can obtain the passing all of user of user
Track data, and excavate the dwell point in user trajectory, position coordinateses including dwell point, user are in the appearance of dwell point
Between, stay time etc..
Alternatively, after the dwell point in user trajectory is determined, first processing units 61 may also be filtered and be wherein
The dwell point of noise types.
Afterwards, dwell point close for geographical position can be clustered by first processing units 61, optionally, for arbitrary poly-
Class result, if including stop points very few, may filter that the cluster result, i.e., the cluster result do not carried out subsequently
Process.
Afterwards, first processing units 61 can be directed to each cluster result, carry out following process respectively:
Using the average of the position coordinateses of each dwell point in the cluster result as the corresponding stroke purpose of the cluster result
The position coordinateses on ground;
The Annual distribution of each dwell point in the cluster result is occurred according to user, determines that the cluster result is corresponding
The day rank regularity weight and all rank regularity weights of stroke destination;
By the day rank regularity weight be compared with corresponding first threshold, by all ranks regularity weight with right
The Second Threshold that answers is compared, if arbitrary weight is more than corresponding threshold value, by corresponding for cluster result stroke destination
It is defined as regular stroke destination, and will be greater than the regularity of the weight as the regular stroke destination of corresponding threshold value
Weight, abandons the weight less than or equal to corresponding threshold value.
For obtaining recent interest stroke destination, second processing unit 62 can obtain user search data, this enforcement first
User search data described in example refer in particular to retrieval data of the user in map, can be determined according to user search data afterwards
The recent interest stroke destination of user.
How user search data are obtained for prior art, second processing unit 62 can obtain the passing all of user of user
Retrieval data.
Alternatively, after user search data are got, the retrieval data wherein for noise types may also be filtered.
Afterwards, the retrieval data of same for correspondence poi can be clustered by second processing unit 62, optionally, for arbitrary
Cluster result, if including retrieval data number very few, may filter that the cluster result, i.e., the cluster result do not carried out
Subsequent treatment.
Afterwards, for each cluster result, second processing unit 62 can carry out following process respectively:
According to retrieval type and the retrieval time of each retrieval data in the cluster result, determine that the cluster result is corresponded to
Poi recent interest weight;
Recent interest weight is compared with corresponding 3rd threshold value, if interest weight is more than the 3rd threshold value in the recent period,
Corresponding for cluster result poi is defined as recent interest stroke destination.
After regular stroke destination and recent interest stroke destination is determined in the manner described above respectively, the 3rd
Processing unit 63 can further determine that out candidate's stroke destination.
Specifically, the 3rd processing unit 63 can be directed to each regular stroke destination, carry out following process respectively:
According to the position coordinateses of the regular stroke destination, determine on map and be located at the regular stroke destination week
Enclose the poi in the region of predefined size;
According to distance and poi temperature with the regular stroke destination, a representative is selected from the poi for determining
Property poi, using the representative poi as candidate's stroke destination, and using the regular weight of the regular stroke destination as
The regular weight of candidate's stroke destination.
3rd processing unit 63 using all recent interest stroke destinatioies all as candidate's stroke destination, and can be incited somebody to action recent
The recent interest weight of interest stroke destination is used as the recent interest weight of candidate's stroke destination.
In addition, the 3rd processing unit 63 also needs to determine the strong time attribute power of each candidate's stroke destination respectively
Weight.
As being directed to each candidate's stroke destination, the 3rd processing unit 63 can obtain retouching for candidate's stroke destination respectively
Information is stated, and by the strong time attribute weight table for pre-setting is inquired about, finds out the description information of candidate's stroke destination
Corresponding strong time attribute weight, using the strong time attribute weight for finding out as candidate's stroke destination strong time attribute
Weight.
When needing to recommend stroke destination to user, it is recommended that unit 64 can determine each candidate's stroke destination respectively
Confidence score, you can for each candidate's stroke destination, respectively according to all weights of candidate's stroke destination, really
Make the confidence score of candidate's stroke destination.
The weight of each candidate's stroke destination may include strong time attribute weight and one below or whole:Regular
Weight, recent interest weight.
First processing units 61 and second processing unit 62 are by regular stroke destination and recent interest stroke purpose
While be sent to the 3rd processing unit 63, can be while that regular weight and recent interest weight are sent to the 3rd process be single
Unit 63, correspondingly, it is recommended that unit 64 can obtain each candidate's stroke destination and corresponding weight from the 3rd processing unit 63, from
And the confidence score of each candidate's stroke destination is calculated respectively according to weight etc..
After the confidence score for respectively obtaining each candidate's stroke destination, it is recommended that unit 64 can according to scoring by height to
Low order is ranked up to each candidate's stroke destination, and the candidate's stroke destination after sequence in front M position is recommended
User, M is that positive integer, concrete value can be decided according to the actual requirements, such as can value be.
The specific workflow of said apparatus embodiment refer to the respective description in preceding method embodiment, herein no longer
Repeat.
In a word, the regular stroke mesh of user, using scheme of the present invention, can be determined according to user trajectory data
Ground, determine the recent interest stroke destination of user according to user search data, and then according to regular stroke destination
Candidate's stroke destination is determined with recent interest stroke destination, so, when needing to recommend stroke destination to user, can
The confidence score of each candidate stroke destination is determined respectively, and according to scoring order from high to low to each candidate's stroke
Destination is ranked up, and then the candidate's stroke destination after sequence in front M position is recommended user, compared to prior art
In only recommend the mode of family and company, can be recommended according to actual demand of user etc. in mode of the present invention, from
And improve the accuracy of recommendation results;And, using scheme of the present invention, can be by regular weight, recent interest power
Weight and strong time attribute weight etc. are determining the confidence score of each candidate's stroke destination, and then are chosen according to appraisal result
Candidate's stroke destination is simultaneously recommended so that recommendation results have very strong time attribute, carry out so as to easily facilitate user
Follow-up stroke planning etc..
In several embodiments provided by the present invention, it should be understood that disclosed apparatus and method, which can be passed through
Its mode is realized.For example, device embodiment described above is only schematically, for example division of the unit, only
Only a kind of division of logic function, can have other dividing mode when actually realizing.
The unit that illustrates as separating component can be or may not be physically separate, aobvious as unit
The part for showing can be or may not be physical location, you can be located at a place, or can also be distributed to multiple
On NE.Some or all of unit therein can be selected according to the actual needs to realize the mesh of this embodiment scheme
's.
In addition, each functional unit in each embodiment of the present invention can be integrated in a processing unit, it is also possible to
It is that unit is individually physically present, it is also possible to which two or more units are integrated in a unit.Above-mentioned integrated list
Unit both can be realized in the form of hardware, it would however also be possible to employ hardware adds the form of SFU software functional unit to realize.
The above-mentioned integrated unit that is realized in the form of SFU software functional unit, can be stored in an embodied on computer readable and deposit
In storage media.Above-mentioned SFU software functional unit is stored in a storage medium, is used so that a computer including some instructions
Equipment (can be personal computer, server, or network equipment etc.) or processor (processor) execute the present invention each
The part steps of embodiment methods described.And aforesaid storage medium includes:USB flash disk, portable hard drive, read only memory (ROM,
Read-Only Memory), random access memory (RAM, Random Access Memory), magnetic disc or CD etc. various
Can be with the medium of store program codes.
Presently preferred embodiments of the present invention is the foregoing is only, not in order to limit the present invention, all essences in the present invention
Within god and principle, any modification, equivalent substitution and improvement that is done etc., should be included within the scope of protection of the invention.
Claims (14)
1. method is recommended by a kind of stroke destination, it is characterised in that include:
The regular stroke destination of user is determined according to user trajectory data, and according to user search data are determined
The recent interest stroke destination of user;
Candidate's stroke destination is determined according to the regularity stroke destination and the recent interest stroke destination;
The confidence score of each candidate stroke destination is determined respectively, and according to scoring order from high to low to each candidate
Stroke destination is ranked up, and the candidate's stroke destination after sequence in front M position is recommended the user, and M is positive integer.
2. method according to claim 1, it is characterised in that
The regular stroke destination for user being determined according to user trajectory data includes:
Obtain the user trajectory data;
By being analyzed to the user trajectory data, the dwell point in the user trajectory is determined;
Filter out temporal regularity from the dwell point often goes to place, used as the regularity stroke destination.
3. method according to claim 2, it is characterised in that
Described filter out temporal regularity from the dwell point often go to place, wrap as the regularity stroke destination
Include:
Dwell point close for geographical position is clustered;
For each cluster result, following process is carried out respectively:
Using the average of the position coordinateses of each dwell point in the cluster result as the corresponding stroke purpose of the cluster result
The position coordinateses on ground;
The Annual distribution of each dwell point in the cluster result is occurred according to the user, determines the cluster result pair
The day rank regularity weight and all rank regularity weights of the stroke destination that answers;
By the day rank regularity weight be compared with corresponding first threshold, by described week rank regularity weight with right
The Second Threshold that answers is compared, if arbitrary weight is more than corresponding threshold value, by corresponding for cluster result stroke purpose
Ground is defined as the regularity stroke destination, and will be greater than the weight of corresponding threshold value as the regularity stroke destination
Regular weight, abandon less than or equal to corresponding threshold value weight.
4. method according to claim 3, it is characterised in that
The user search data are retrieval data of the user in map;
The recent interest stroke destination for the user being determined according to user search data includes:
The retrieval data of same for correspondence point of interest poi are clustered;
For each cluster result, following process is carried out respectively:
According to retrieval type and the retrieval time of each retrieval data in the cluster result, determine that the cluster result is corresponded to
Poi recent interest weight;
The recent interest weight is compared with corresponding 3rd threshold value, if the recent interest weight is more than the described 3rd
Threshold value, then be defined as the recent interest stroke destination by corresponding for cluster result poi.
5. method according to claim 4, it is characterised in that
Described according to described regularity stroke destination and the recent interest stroke destination determine candidate's stroke destination
Including:
For each regular stroke destination, following process is carried out respectively:
According to the position coordinateses of the regularity stroke destination, determine on map and be located at regularity stroke destination week
Enclose the poi in the region of predefined size;
According to distance and poi temperature with the regularity stroke destination, a representativeness is selected from the poi for determining
Poi, using the representativeness poi as candidate's stroke destination, and the regular power by the regularity stroke destination
Recast is the regular weight of candidate's stroke destination;
Using the recent interest stroke destination as candidate's stroke destination, and by the recent interest stroke destination
Recent interest weight as candidate's stroke destination recent interest weight.
6. method according to claim 5, it is characterised in that
Before the confidence score for determining each candidate's stroke destination respectively, further include:Determine respectively every
The strong time attribute weight of individual candidate's stroke destination;
The confidence score for determining each candidate's stroke destination respectively includes:For each candidate's stroke destination,
Respectively according to all weights of candidate's stroke destination, the confidence score of candidate's stroke destination is determined.
7. method according to claim 6, it is characterised in that
The strong time attribute weight for determining each candidate's stroke destination respectively includes:
For each candidate's stroke destination, the description information of candidate's stroke destination is obtained respectively, and pre- by inquiry
The strong time attribute weight table for first arranging, finds out the corresponding strong time attribute power of description information of candidate's stroke destination
Weight, using the strong time attribute weight for finding out as candidate's stroke destination strong time attribute weight.
8. a kind of stroke destination recommendation apparatus, it is characterised in that include:First processing units, second processing unit, at the 3rd
Reason unit and recommendation unit;
The first processing units, for determining the regular stroke destination of user according to user trajectory data, and send
To the 3rd processing unit;
The second processing unit, for determining the recent interest stroke destination of the user according to user search data,
And it is sent to the 3rd processing unit;
3rd processing unit, for determining according to the regularity stroke destination and the recent interest stroke destination
Go out candidate's stroke destination;
The recommendation unit, for determining the confidence score of each candidate's stroke destination respectively, and according to scoring by height
To low order, each candidate's stroke destination is ranked up, the candidate's stroke destination after sequence in front M position is recommended
The user, M is positive integer.
9. device according to claim 8, it is characterised in that
The first processing units obtain the user trajectory data;By being analyzed to the user trajectory data, determine
The dwell point for going out in the user trajectory;Filter out temporal regularity from the dwell point often goes to place, used as the rule
Lv Xing stroke destination.
10. device according to claim 9, it is characterised in that
Dwell point close for geographical position is clustered by the first processing units;
For each cluster result, following process is carried out respectively:
Using the average of the position coordinateses of each dwell point in the cluster result as the corresponding stroke purpose of the cluster result
The position coordinateses on ground;
The Annual distribution of each dwell point in the cluster result is occurred according to the user, determines the cluster result pair
The day rank regularity weight and all rank regularity weights of the stroke destination that answers;
By the day rank regularity weight be compared with corresponding first threshold, by described week rank regularity weight with right
The Second Threshold that answers is compared, if arbitrary weight is more than corresponding threshold value, by corresponding for cluster result stroke purpose
Ground is defined as the regularity stroke destination, and will be greater than the weight of corresponding threshold value as the regularity stroke destination
Regular weight, abandon less than or equal to corresponding threshold value weight.
11. devices according to claim 10, it is characterised in that
The user search data are retrieval data of the user in map;
The retrieval data of same for correspondence point of interest poi are clustered by the second processing unit;
For each cluster result, following process is carried out respectively:
According to retrieval type and the retrieval time of each retrieval data in the cluster result, determine that the cluster result is corresponded to
Poi recent interest weight;
The recent interest weight is compared with corresponding 3rd threshold value, if the recent interest weight is more than the described 3rd
Threshold value, then be defined as the recent interest stroke destination by corresponding for cluster result poi.
12. devices according to claim 11, it is characterised in that
3rd processing unit is directed to each regular stroke destination, carries out following process respectively:
According to the position coordinateses of the regularity stroke destination, determine on map and be located at regularity stroke destination week
Enclose the poi in the region of predefined size;
According to distance and poi temperature with the regularity stroke destination, a representativeness is selected from the poi for determining
Poi, using the representativeness poi as candidate's stroke destination, and the regular power by the regularity stroke destination
Recast is the regular weight of candidate's stroke destination;
Using the recent interest stroke destination as candidate's stroke destination, and by the recent interest stroke destination
Recent interest weight as candidate's stroke destination recent interest weight.
13. devices according to claim 12, it is characterised in that
3rd processing unit is further used for, and determines the strong time attribute weight of each candidate's stroke destination respectively;
The recommendation unit is directed to each candidate's stroke destination, respectively according to all weights of candidate's stroke destination,
Determine the confidence score of candidate's stroke destination.
14. devices according to claim 13, it is characterised in that
3rd processing unit is directed to each candidate's stroke destination, obtains the description letter of candidate's stroke destination respectively
Breath, and by the strong time attribute weight table for pre-setting is inquired about, find out the description information pair of candidate's stroke destination
The strong time attribute weight that answers, using the strong time attribute weight for finding out as candidate's stroke destination strong time attribute
Weight.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610844639.7A CN106446157B (en) | 2016-09-22 | 2016-09-22 | Travel destination recommendation method and device |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610844639.7A CN106446157B (en) | 2016-09-22 | 2016-09-22 | Travel destination recommendation method and device |
Publications (2)
Publication Number | Publication Date |
---|---|
CN106446157A true CN106446157A (en) | 2017-02-22 |
CN106446157B CN106446157B (en) | 2020-01-21 |
Family
ID=58167129
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610844639.7A Active CN106446157B (en) | 2016-09-22 | 2016-09-22 | Travel destination recommendation method and device |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106446157B (en) |
Cited By (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110020111A (en) * | 2017-09-21 | 2019-07-16 | 腾讯科技(深圳)有限公司 | Stroke recommended method, device, computer equipment and storage medium |
CN110059260A (en) * | 2019-04-29 | 2019-07-26 | 北京字节跳动网络技术有限公司 | A kind of recommended method, device, equipment and medium |
CN110516017A (en) * | 2019-08-02 | 2019-11-29 | Oppo广东移动通信有限公司 | Location information processing method, device, electronic equipment and storage medium based on terminal device |
CN111611500A (en) * | 2020-04-09 | 2020-09-01 | 中国平安财产保险股份有限公司 | Frequent place identification method and device based on clustering and storage medium |
WO2021087663A1 (en) * | 2019-11-04 | 2021-05-14 | Beijing Didi Infinity Technology And Development Co., Ltd. | Systems and methods for determining name for boarding point |
CN113590674A (en) * | 2021-06-29 | 2021-11-02 | 北京百度网讯科技有限公司 | Travel purpose identification method, device, equipment and storage medium |
US11710142B2 (en) | 2018-06-11 | 2023-07-25 | Beijing Didi Infinity Technology And Development Co., Ltd. | Systems and methods for providing information for online to offline service |
Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103542848A (en) * | 2012-07-17 | 2014-01-29 | 株式会社电装 | Destination recommendation system and destination recommendation method |
US20140149035A1 (en) * | 2012-11-23 | 2014-05-29 | Hyundai Mnsoft, Inc. | Method and device for generating intersection guidance information |
CN104102719A (en) * | 2014-07-18 | 2014-10-15 | 百度在线网络技术(北京)有限公司 | Track information pushing method and device |
CN104410717A (en) * | 2014-12-18 | 2015-03-11 | 百度在线网络技术(北京)有限公司 | Information pushing method and device |
CN104520881A (en) * | 2012-06-22 | 2015-04-15 | 谷歌公司 | Ranking nearby destinations based on visit likelihoods and predicting future visits to places from location history |
CN104899252A (en) * | 2015-05-12 | 2015-09-09 | 北京嘀嘀无限科技发展有限公司 | Information push method and apparatus |
CN105183800A (en) * | 2015-08-25 | 2015-12-23 | 百度在线网络技术(北京)有限公司 | Information prediction method and apparatus |
-
2016
- 2016-09-22 CN CN201610844639.7A patent/CN106446157B/en active Active
Patent Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104520881A (en) * | 2012-06-22 | 2015-04-15 | 谷歌公司 | Ranking nearby destinations based on visit likelihoods and predicting future visits to places from location history |
CN103542848A (en) * | 2012-07-17 | 2014-01-29 | 株式会社电装 | Destination recommendation system and destination recommendation method |
US20140149035A1 (en) * | 2012-11-23 | 2014-05-29 | Hyundai Mnsoft, Inc. | Method and device for generating intersection guidance information |
CN104102719A (en) * | 2014-07-18 | 2014-10-15 | 百度在线网络技术(北京)有限公司 | Track information pushing method and device |
CN104410717A (en) * | 2014-12-18 | 2015-03-11 | 百度在线网络技术(北京)有限公司 | Information pushing method and device |
CN104899252A (en) * | 2015-05-12 | 2015-09-09 | 北京嘀嘀无限科技发展有限公司 | Information push method and apparatus |
CN105183800A (en) * | 2015-08-25 | 2015-12-23 | 百度在线网络技术(北京)有限公司 | Information prediction method and apparatus |
Cited By (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110020111A (en) * | 2017-09-21 | 2019-07-16 | 腾讯科技(深圳)有限公司 | Stroke recommended method, device, computer equipment and storage medium |
CN110020111B (en) * | 2017-09-21 | 2023-02-24 | 腾讯科技(深圳)有限公司 | Travel recommendation method and device, computer equipment and storage medium |
US11710142B2 (en) | 2018-06-11 | 2023-07-25 | Beijing Didi Infinity Technology And Development Co., Ltd. | Systems and methods for providing information for online to offline service |
CN110059260A (en) * | 2019-04-29 | 2019-07-26 | 北京字节跳动网络技术有限公司 | A kind of recommended method, device, equipment and medium |
CN110059260B (en) * | 2019-04-29 | 2020-07-31 | 北京字节跳动网络技术有限公司 | Recommendation method, device, equipment and medium |
CN110516017A (en) * | 2019-08-02 | 2019-11-29 | Oppo广东移动通信有限公司 | Location information processing method, device, electronic equipment and storage medium based on terminal device |
WO2021087663A1 (en) * | 2019-11-04 | 2021-05-14 | Beijing Didi Infinity Technology And Development Co., Ltd. | Systems and methods for determining name for boarding point |
CN111611500A (en) * | 2020-04-09 | 2020-09-01 | 中国平安财产保险股份有限公司 | Frequent place identification method and device based on clustering and storage medium |
CN113590674A (en) * | 2021-06-29 | 2021-11-02 | 北京百度网讯科技有限公司 | Travel purpose identification method, device, equipment and storage medium |
Also Published As
Publication number | Publication date |
---|---|
CN106446157B (en) | 2020-01-21 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106446157A (en) | Route destination recommending method and device | |
CN103646069B (en) | Recommendation method and device for ticket service information | |
CN107798412B (en) | Route recommendation method and device | |
CN104978420B (en) | Traffic route matching process and device | |
US9208511B2 (en) | System and method for location-based recommendations | |
CN104991924B (en) | Method and apparatus for the address for determining new supply centre | |
JP4213199B1 (en) | Information provision system | |
US9779143B2 (en) | Information pushing method and apparatus | |
CN108108821A (en) | Model training method and device | |
US9857194B2 (en) | Time related points of interest for navigation system | |
KR20180048788A (en) | Methods and apparatuses for determining the need for placement on points of interest | |
CN106557474A (en) | Obtain the method and device of POI, database, navigation terminal and automobile on the way | |
JP6756744B2 (en) | Location information provision method and equipment | |
CN105865478A (en) | Navigation information pushing method, and apparatus and device thereof | |
US20160034968A1 (en) | Method and device for determining target user, and network server | |
CN105975537A (en) | Sorting method and device of application program | |
US20140379476A1 (en) | Method and data processing apparatus | |
CN105160019B (en) | Card display methods and device | |
CN111831899B (en) | Navigation interest point recommendation method, device, server and readable storage medium | |
KR20200003109A (en) | Method and apparatus for setting sample weight, electronic device | |
CN106248096B (en) | The acquisition methods and device of road network weight | |
WO2016123867A1 (en) | Method and device for position search cognition | |
WO2016175940A1 (en) | Determining semantic place names from location reports | |
CN107133689B (en) | Position marking method | |
CN106503071A (en) | The processing method and processing device of POI |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |