CN113742608A - Model training method, destination recommendation method, and storage medium - Google Patents
Model training method, destination recommendation method, and storage medium Download PDFInfo
- Publication number
- CN113742608A CN113742608A CN202110844528.7A CN202110844528A CN113742608A CN 113742608 A CN113742608 A CN 113742608A CN 202110844528 A CN202110844528 A CN 202110844528A CN 113742608 A CN113742608 A CN 113742608A
- Authority
- CN
- China
- Prior art keywords
- historical
- starting
- travel
- departure
- 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.)
- Pending
Links
- 238000012549 training Methods 0.000 title claims abstract description 82
- 238000000034 method Methods 0.000 title claims abstract description 73
- 239000013598 vector Substances 0.000 claims description 18
- 238000013507 mapping Methods 0.000 claims description 7
- 230000004927 fusion Effects 0.000 claims description 6
- 238000012163 sequencing technique Methods 0.000 claims description 6
- 238000006243 chemical reaction Methods 0.000 claims description 5
- 238000004590 computer program Methods 0.000 claims description 5
- 230000001960 triggered effect Effects 0.000 claims description 3
- 230000000875 corresponding effect Effects 0.000 description 79
- 238000010586 diagram Methods 0.000 description 8
- 230000006870 function Effects 0.000 description 5
- 230000000694 effects Effects 0.000 description 4
- 230000006399 behavior Effects 0.000 description 3
- 238000013145 classification model Methods 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 230000002776 aggregation Effects 0.000 description 2
- 238000004220 aggregation Methods 0.000 description 2
- 230000001174 ascending effect Effects 0.000 description 2
- 238000004422 calculation algorithm Methods 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 238000007667 floating Methods 0.000 description 2
- 230000003993 interaction Effects 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 238000012935 Averaging Methods 0.000 description 1
- 230000002159 abnormal effect Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005587 bubbling Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 230000002596 correlated effect Effects 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000008713 feedback mechanism Effects 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000007500 overflow downdraw method Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Images
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
- G06F16/9537—Spatial or temporal dependent retrieval, e.g. spatiotemporal queries
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/29—Geographical information databases
-
- 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
- G06F16/906—Clustering; Classification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
- G06F16/9535—Search customisation based on user profiles and personalisation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/40—Business processes related to the transportation industry
Landscapes
- Engineering & Computer Science (AREA)
- Databases & Information Systems (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Remote Sensing (AREA)
- Health & Medical Sciences (AREA)
- Economics (AREA)
- General Health & Medical Sciences (AREA)
- Human Resources & Organizations (AREA)
- Marketing (AREA)
- Primary Health Care (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The embodiment of the application provides a model training method, a destination recommendation method and a storage medium. The method comprises the following steps: generating a query request based on a first historical travel record of a travel object; recalling historical starting and ending point pairs of the travel object based on the query request; based on the starting features corresponding to the historical starting and ending point pairs, constructing training samples corresponding to the historical starting and ending point pairs, wherein the training samples comprise: similarity characteristics used for representing the similarity between the starting characteristics corresponding to the historical starting and ending point pairs and the query request; marking training labels of corresponding training samples of the historical starting and ending points according to the real destination corresponding to the query request and the ending point in the historical starting and ending points; and training a preset destination recommendation model according to the marked training samples. The technical scheme provided by the embodiment of the application has higher recommendation accuracy and can improve the trip efficiency of the trip object.
Description
Technical Field
The present application relates to the field of computer technologies, and in particular, to a model training method, a destination recommendation method, and a storage medium.
Background
With the rapid development of the internet and intelligent terminal technology, people can use software which is installed on intelligent terminals such as mobile phones and supports the network car booking function to meet the self travel requirement. When the trip object uses the software to make a taxi, the destination can be manually searched in the search box.
In order to facilitate the travel of a travel object, reduce the operation and improve the trip experience of the travel object, some software can actively recommend a travel place (such as a destination) for the travel object when the software is opened by the travel object at present, and whether the recommended destination meets the expectation of the travel object or not can directly influence the travel experience of the travel object. Therefore, how to ensure that the actively recommended destination meets the travel object expectation is a problem that needs to be continuously solved and optimized by those skilled in the art.
Disclosure of Invention
In view of the above, the present application is proposed to provide a model training method, a destination recommendation method, and a storage medium that solve the above problems or at least partially solve the above problems.
Thus, in one embodiment of the present application, a model training method is provided. The method comprises the following steps:
generating a query request based on a first historical travel record of a travel object;
recalling historical starting and ending point pairs of the travel object based on the query request;
based on the starting features corresponding to the historical starting and ending point pairs, constructing training samples corresponding to the historical starting and ending point pairs, wherein the training samples comprise: similarity characteristics used for representing the similarity between the starting characteristics corresponding to the historical starting and ending point pairs and the query request;
marking training labels of corresponding training samples of the historical starting and ending points according to the real destination corresponding to the query request and the ending point in the historical starting and ending points;
and training a preset destination recommendation model according to the marked training samples.
In yet another embodiment of the present application, a destination recommendation method is provided. The method comprises the following steps:
acquiring a query request of a trip object;
recalling historical starting and ending point pairs of the travel object based on the query request;
determining input information based on the departure characteristics corresponding to the historical start and end points, wherein the input information comprises: similarity characteristics used for representing the similarity between the starting characteristics corresponding to the historical starting and ending point pairs and the query request;
inputting the input information into a trained destination recommendation model to obtain a prediction result;
and determining whether the terminal point in the historical starting and terminal point pair is a destination to be recommended to the travel object according to the prediction result. In another embodiment of the present application, a computer-readable storage medium storing a computer program is provided, wherein the computer program is capable of implementing any one of the model training method or the destination recommendation method described above when executed by a computer.
In the technical scheme provided by the embodiment of the application, the greater the similarity between the starting feature corresponding to the historical starting and ending point pair of the travel object and the query request of the travel object, the greater the probability that the travel object goes to the ending point in the historical starting and ending point pair. Then, a training sample for training the destination recommendation model is constructed according to the similarity characteristic of the similarity between the starting characteristic corresponding to the historical starting and ending point pair representing the travel object and the query request of the travel object, and the recommendation accuracy of the destination recommendation model obtained through training is improved.
In the technical scheme provided by the embodiment of the application, the greater the similarity between the starting feature corresponding to the historical starting and ending point pair of the travel object and the query request of the travel object, the greater the probability that the travel object goes to the ending point in the historical starting and ending point pair. Then, the similarity characteristic of the similarity between the starting characteristic corresponding to the historical starting and ending point pair of the travel object and the query request of the travel object is input into the destination recommendation model, so that the recommendation accuracy of the destination recommendation model is improved, and the travel efficiency of the travel object can be improved.
Drawings
In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings needed to be used in the description of the embodiments or the prior art will be briefly described below, and it is obvious that the drawings in the following description are some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative efforts.
Fig. 1 is a schematic flowchart of a model training method according to an embodiment of the present application;
fig. 2 is a schematic flowchart of a destination recommendation method according to an embodiment of the present application;
FIG. 3 is a diagram of an example of an interface according to an embodiment of the present application;
FIG. 4 is a diagram of a second example of an interface provided in an embodiment of the present application;
fig. 5 is a schematic flowchart of a human-computer interaction method according to an embodiment of the present application;
FIG. 6 is a block diagram of a model training apparatus according to an embodiment of the present disclosure;
fig. 7 is a block diagram of a destination recommendation apparatus according to another embodiment of the present application;
fig. 8 is a block diagram of an electronic device according to another embodiment of the present application.
Detailed Description
At present, a common destination recommendation scheme is mainly policy-based recommendation, that is, a plurality of destinations that a travel object has arrived from an order starting point are recalled according to the order starting point of the travel object history; and then, arranging the plurality of destinations in a descending order according to the heat degree, and taking the first destination as a recommended destination.
The method for recalling the candidate destination according to the order starting point of the travel object has the problem of low coverage rate of the recalled destination, so that the recommendation accuracy rate is reduced; further, determining the recommended destination only based on the degree of heat of the destination further reduces the accuracy of recommendation. Therefore, the application provides a new destination recommendation method. According to the method, the similarity characteristic of the similarity between the starting characteristic corresponding to the historical starting and ending point pair of the travel object and the query request of the travel object is used as the prediction basis of the destination recommendation model, so that the model prediction accuracy is improved, the destination recommendation accuracy is improved, and the user experience is improved.
In order to make the technical solutions better understood by those skilled in the art, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. It is to be understood that the embodiments described are only a few embodiments of the present application and not all embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.
Further, in some flows described in the specification, claims, and above-described figures of the present application, a number of operations are included that occur in a particular order, which operations may be performed out of order or in parallel as they occur herein. The sequence numbers of the operations, e.g., 101, 102, etc., are used merely to distinguish between the various operations, and do not represent any order of execution per se. Additionally, the flows may include more or fewer operations, and the operations may be performed sequentially or in parallel. It should be noted that, the descriptions of "first", "second", etc. in this document are used for distinguishing different messages, devices, modules, etc., and do not represent a sequential order, nor limit the types of "first" and "second" to be different.
Fig. 1 shows a schematic flow chart of a model training method according to an embodiment of the present application. The execution main body of the method can be a client or a server. The client may be hardware integrated on the terminal and having an embedded program, may also be application software installed in the terminal, and may also be tool software embedded in an operating system of the terminal, which is not limited in this embodiment of the present application. The terminal can be any terminal equipment including a mobile phone, a tablet computer, a vehicle-mounted terminal equipment and the like. The server may be a common server, a cloud, a virtual server, or the like, which is not specifically limited in this embodiment of the application. As shown in fig. 1, the method includes:
101. and generating a query request based on the first historical travel record of the travel object.
102. And recalling historical starting and ending point pairs of the travel object based on the query request.
103. And constructing training samples corresponding to the historical starting and ending point pairs based on the starting features corresponding to the historical starting and ending point pairs.
Wherein the training samples comprise: and the similarity characteristic is used for representing the similarity between the starting characteristic corresponding to the historical starting and ending point pair and the query request.
104. And marking the training labels of the training samples corresponding to the historical starting and ending points according to the real destination corresponding to the query request and the ending points in the historical starting and ending points.
105. And training a preset destination recommendation model according to the marked training samples.
In the above 101, the first historical travel record may be determined from the historical travel record of the travel object, and in an example, the first historical travel record may be randomly determined from the historical travel record of the travel object. In another example, the last historical travel record of the travel object may be determined as the first historical travel record. The determination manner of the first historical travel record may be set according to actual needs, and embodiments of the present application are not particularly limited.
The first historical travel record may include, but is not limited to, object identification information, travel area information, and departure information. The departure information may include a departure location, a departure time period, and/or a number of departure weeks, among others. Wherein the object identification information may include a travel object ID number. The trip area information refers to geographical area information related to the first historical trip record. The travel area information may include travel area identification, such as: and (5) city coding.
In an example, the query request may carry object identification information, travel area information, and/or departure information of the first historical travel record.
In the above 102, the historical starting and ending point pair of the travel object indicates that the travel object has historically arrived from the starting point in the historical starting and ending point to the ending point in the historical starting and ending point pair.
In one example, the historical starting and ending point pairs of the travel object can be recalled based on object identification information carried in the query request. In this embodiment, the recalled historical starting and ending point pairs of the travel object may be understood as all historical starting and ending point pairs on the travel object history.
In another example, the historical starting and ending point pairs of the travel object can be recalled based on the object identification information and the travel area information carried in the query request. In this embodiment, the recalled historical starting and ending point pairs of the travel object may be understood as all historical starting and ending point pairs of the travel object in the travel area historically. Specifically, the above-mentioned "recalling the historical starting and ending point pairs of the travel object based on the query request" in 102 may include:
1021. and recalling historical starting and ending point pairs of the travel object related to the historical travel area based on the object identification information and the travel area information of the travel object carried in the query request.
In 103, the similarity feature used to represent the similarity between the departure feature corresponding to the historical starting and ending point pair and the query request is specifically a similarity feature used to represent the similarity between the departure feature corresponding to the historical starting and ending point pair and the travel information carried in the query request.
The starting characteristics corresponding to the historical starting and ending point pairs are used for characterizing the characteristics of the travel information of the travel object about at least one second historical travel record of the historical starting and ending point pairs.
In an example, the first time period in which the departure time is located may also be determined according to the departure time carried in the query request. Recalling historical starting and ending point pairs of the travel object related to the historical travel area based on the object identification information and the travel area information of the travel object carried in the query request, specifically comprising: and recalling historical starting and ending points of the travel object in the historical travel area within the first time period based on the object identification information, the travel area information and the first time period of the travel object carried in the query request.
The determination method of the similarity characteristic of the similarity between the departure characteristic and the travel information will be described in detail in the following embodiments.
In the step 104, the real destination corresponding to the query request is determined according to the first historical travel record. Specifically, the end point in the first historical travel record may be used as the real destination corresponding to the query request.
And marking the training labels of the training samples corresponding to the historical starting and ending points according to the real destination corresponding to the query request and the ending points in the historical starting and ending points.
In an example, when an end point in the historical start and end points is the same as the real destination, the training label of the historical start and end point to the corresponding training sample can be determined to be 1; when the end point in the historical start and end points is different from the real destination, the training label of the corresponding training sample of the historical start and end point pair can be determined as 0.
In this embodiment, the destination recommendation model is a classification model.
In another example, the history starting and ending points are multiple. 1034, marking the training label of the training sample corresponding to the history starting and ending point according to the real destination corresponding to the query request and the ending point in the history starting and ending point specifically may be implemented by:
1041. and sequencing the plurality of historical starting and ending point pairs according to the distances between the real destination corresponding to the query request and the ending points in the plurality of historical starting and ending points respectively.
1042. And respectively marking the training labels of the training samples corresponding to the plurality of historical starting and ending point pairs according to the sequencing results of the plurality of historical starting and ending point pairs.
Illustratively, in 1041, the plurality of history starting and ending point pairs are sorted in ascending order according to the distance between the real destination and the ending point of the plurality of history starting and ending points, respectively.
For example, in the above 1042, the ranking numbers of the plurality of history starting and ending point pairs may be respectively used as the training labels of the training samples corresponding to the plurality of history starting and ending point pairs.
For example: the plurality of history starting and ending point pairs comprise O1D1, O2D2 and O3D3, wherein the distances between the real destination D and the D1, the D2 and the D3 are 1km, 2km and 3km in sequence. And sorting the obtained O1D1, O2D2 and O3D3 in ascending order according to the distance, wherein the sorting sequence numbers are 1, 2 and 3 respectively. Then, the training labels of the training samples corresponding to O1D1, O2D2, and O3D3 are 1, 2, and 3 in sequence.
In a preferred implementation manner of this embodiment, the destination recommendation model is a ranking model. Because the sorting model can take the sorting relation among a plurality of historical starting and ending point pairs into account, and the classification model cannot take the sorting relation among the historical starting and ending point pairs into account. Therefore, the use of the ranking model is a preferred embodiment that can further improve the prediction accuracy of the model, and the classification model is a less preferred embodiment.
In the above 105, a preset destination recommendation model is trained according to the labeled training samples.
In an example, the marked training samples may be input into a preset destination recommendation model, so as to obtain a prediction result of the destination recommendation model. And optimizing the destination recommendation model according to the prediction result and the label of the training sample.
A loss function may be employed to calculate a loss value between the prediction result and the labels of the training samples; the model is optimized according to the loss value. Specifically, the input of the loss function may be determined according to the determination result and the label of the training sample. The loss function may be selected according to actual needs, which is not specifically limited in the embodiment of the present application. The specific optimization process can be found in the prior art, and is not described in detail herein.
In the technical scheme provided by the embodiment of the application, the greater the similarity between the starting feature corresponding to the historical starting and ending point pair of the travel object and the query request of the travel object, the greater the probability that the travel object goes to the ending point in the historical starting and ending point pair. Then, a training sample for training the destination recommendation model is constructed according to the similarity characteristic of the similarity between the starting characteristic corresponding to the historical starting and ending point pair representing the travel object and the query request of the travel object, and the recommendation accuracy of the destination recommendation model obtained through training is improved.
The destination recommendation model is a recommendation model Based on deep learning and location Based services (lbs). The model recalls the destination through a Candidate Network (Candidate Generation Network), and a better recalling effect can be obtained. In an example, the destination recommendation model may be a Gradient lifting tree model, and specifically may be an XGBoost (eXtreme Gradient lifting) model. By adopting the model, the model expansion capability and the fitting capability are greatly improved, so that the model can utilize richer features to recommend the destination, and the defects of low precision, coverage and high error rate of a strategy recommendation scheme can be overcome. In addition, the sorting effect of the integrated tree model after distributed training based on the mass samples is far better than that of a strategy recommendation scheme, and the sorting effect is better.
The following describes the determination manner of the historical starting and ending points to the corresponding starting characteristics:
the method further comprises the following steps:
106. and acquiring more than one second historical travel record of the travel object.
And the historical departure positions of all the second historical travel records are located in the first geographic grid, and the historical destination positions are located in the second geographic grid.
107. And determining the starting characteristics corresponding to the historical starting and ending point pairs according to the historical starting information of the second historical travel record.
The starting point of the historical starting and end point pair comprises a first geographic grid, and the end point comprises a second geographic grid.
The recording time of all the second historical travel records is earlier than that of the first historical travel record, so that true values can be prevented from being leaked during model training. All the second historical travel records may be historical travel records within a preset time period. For example: the preset time period may be within the last 90 days.
The departure characteristics are used for characterizing the historical departure information of more than one second historical travel record, and may include one or more of the following items: departure time characteristic, departure position characteristic, departure week number characteristic.
In an implementation manner, in 106, at least one piece of second historical departure information, in which the historical departure position of the travel object is located in the first geographic grid, the historical destination position is located in the second geographic grid, and the historical departure time is located in the first time period, may be obtained.
The time periods can be divided in advance for one day to obtain a plurality of time periods. Specifically, 24 hours per day may be divided into four time periods of an early peak, a daytime peak, a late peak, and a night peak based on road congestion conditions. Therefore, the time period of the trip is determined according to the historical departure time in the historical trip record.
In the embodiment, the training samples are established in different time periods, which is helpful for improving the accuracy of the model. This is because, for the same trip subject, there may be differences in destinations during different time periods even if their starting points are the same.
In practical application, the following steps can be adopted to determine a historical starting and ending point pair:
108. and acquiring a third history travel record of the travel object.
109. And determining the geographic grid to which the historical departure position of the third historical travel record belongs as the first geographic grid.
110. And determining the geographical grid to which the historical target position of the third historical travel record belongs to serve as the second geographical grid.
The third historical travel record is different from the first historical travel record. In practical application, the recording time of the third historical travel record is earlier than that of the first historical travel record.
The longitude and latitude of the historical departure position in the third historical travel record can be converted into a first h3 index with first preset precision; and taking the corresponding grid indexed by the first h3 as a first geographic grid. The longitude and latitude of the historical target position in the third historical travel record can be converted into a second h3 index with second preset precision; and taking the corresponding grid indexed by the second h3 as a second geographic grid. The first preset precision and the second preset precision can be set according to actual needs, and this is not specifically limited in the embodiments of the present application.
The higher the precision is, the smaller the grid corresponding to the h3 index is, and the smaller the geographical area covered by the grid is. The technology related to h3 indexing and its implementation principle can be referred to in the prior art, and will not be described in detail herein. In one example, the geographic area covered by the grid corresponding to the h3 index may be 1km2-2km2In the meantime. By adopting the geographic grid, the historical travel records of the travel object can be aggregated, so that the aggregation characteristics (namely the starting characteristics) can be determined, and the accuracy can be improved by recommending the aggregation characteristics.
In this embodiment, a first geographic grid is determined, that is, a starting point of a historical starting and ending point pair is determined; a second geographic grid is determined, i.e., the endpoints of the historical starting and ending point pairs are determined.
In a specific example, the historical departure information includes a historical departure location. The step 107 of "determining the departure characteristics corresponding to the historical starting and ending points according to the historical departure information of the second historical travel record" may include:
1071. and fusing the historical starting positions of the second historical travel record to obtain a fused historical starting position which is used as one of the starting characteristics corresponding to the historical starting and ending point pairs.
And the similarity feature comprises a position similarity feature used for representing the similarity between the fused historical departure position and the departure position carried in the query request. The position similarity feature may specifically be a linear distance and a manhattan distance between the fused historical departure position and the departure position carried in the query request.
Since the longitude and latitude of the historical departure position in the at least one second historical travel record are floating point numbers and may be repeated for many times, the departure positions need to be aggregated and then the average value is calculated. And fusing the historical departure positions of more than one second historical trip record, namely searching the cluster center position in the historical departure positions of more than one second historical trip record to serve as the fused historical departure position.
Due to the fused historical departure positions, the system is extremely prone to being interfered by abnormal points, such as the historical departure positions p1, p2, p9 are all at the position A, and the historical departure position p3 is at the position B which is 3 kilometers away. If the longitude and latitude of the historical departure position are averaged, the connection line 1/2 in the AB is taken as the clustering center position, which is obviously unreasonable. Moreover, the clustering center position finally determined by the existing clustering algorithm is usually not a real starting position, which may affect the calculation rationality of the similarity feature to a certain extent.
Therefore, the applicant proposes a new location fusion method, specifically, in the above 1071, "fusion of historical departure locations in the plurality of second historical travel records to obtain a fused historical departure location" may be implemented by adopting the following steps:
and S11, determining the radius of a minimum enclosing circle taking the historical departure positions as the center of the circle for each historical departure position in the historical departure positions of the second historical travel record.
Wherein the minimum bounding circle refers to a minimum circle that bounds at least a set number of the historical departure positions; the set number is equal to the product of the number of the second historical travel records and a set preset proportional coefficient.
And S12, determining the center of the minimum enclosing circle with the minimum radius as the history starting position after fusion.
In S11, the radius of the smallest enclosing circle around each historical departure position may be determined by a half mass radius (half radius) algorithm.
In this example, the finally determined clustering center position (i.e., the fused historical departure position) is the real departure position on the travel object history, which is helpful for improving the calculation rationality of the similarity characteristics, so as to improve the recommendation accuracy of the model.
In another specific example, the historical departure information includes historical departure time. Since the historical departure time in the at least one second historical travel record is a floating point number and may be repeated for a plurality of times, the departure times need to be aggregated and then averaged. In 107, the "determining the departure characteristics corresponding to the historical starting and ending points according to the historical departure information of the second historical travel record" may further include:
1072. and fusing the historical departure time of the second historical travel record to obtain fused historical departure time which is used as one of the departure characteristics corresponding to the historical starting and ending point pairs.
And the similarity feature comprises a time similarity feature used for representing the similarity between the fused historical departure time and the departure time carried in the query request. The time similarity feature may specifically be a difference between the fused historical departure time and the departure time carried in the query request.
And fusing the historical departure time of more than one second historical travel record, namely determining the average value of the historical departure time to serve as the fused historical departure time.
Since it is time that is a circular quantity, it is always cyclically varied within 0-24. If the arithmetic average of the departure time of the at least one second historical travel record is taken as the historical departure time average, unreasonable situations may occur: for example, 1 point and 23 points, and 24 points or 0 points in the middle are reasonably suitable as the average value of the historical departure time; however, it is not reasonable to set the arithmetic mean 12 of the two as the historical departure time mean. Therefore, the applicant proposes to vectorize time and determine the historical departure time average by calculating the vector average. Specifically, 1072 "merge the historical departure time of the second historical travel record to obtain a merged historical departure time as one of the departure characteristics corresponding to the historical starting and ending point pair", may specifically be implemented by:
and S21, mapping the historical departure time of the second historical trip record into angles under a polar coordinate system respectively.
And S22, obtaining a unit vector under a rectangular coordinate system through coordinate conversion according to the angle.
And S23, obtaining the target angle in the polar coordinate system through coordinate conversion according to the mean vector of the unit vector.
And S24, mapping the target angle to target time to be used as the history departure time after fusion.
In S21, the polar coordinate has an angle range of 0 to 360 degrees, and a corresponding time range of 0 to 24 h. The corresponding mapping formula (1) is specifically:
wherein, thetaiRefers to the departure time t in the ith second historical travel record in more than one second historical travel recordiAnd mapping to an angle under a polar coordinate system.
In the above-mentioned S22, θiThrough the transformation of the coordinates, the system can be used,the unit vector under the rectangular coordinate system is obtained as (cos theta)i，sinθi)。
In S23, the mean vector of at least one unit vector is:
wherein n is the total number of at least one unit vector, that is, the total number of the second historical travel records.
Mean vector
Obtaining the reference angle under the polar coordinate system through coordinate conversion
The corresponding formula is as follows:
at S14, the reference angle may be mapped to time by using the following mapping formula (3):
the time of averaging 1 point and 23 points by adopting the vector method is 0 point, and obviously, the time average value obtained by the vector method is more reasonable.
In another specific example, the historical starting and ending point pairs corresponding to the starting features include the starting week number of the last trip record in the second historical trip record. The similarity feature comprises a week similarity feature used for representing similarity between the departure week number of the last trip record in the second historical trip record and the departure week number carried in the query request. The week number similarity characteristic may specifically be a difference between the departure week number of the last trip record in the second historical trip record and the departure week number carried in the query request.
In the embodiment of the application, the query request and the historical travel record of the travel object are not directly input into the model, but are preprocessed, that is, the similarity characteristic used for representing the similarity between the departure characteristic corresponding to the historical start and end point pair and the query request is determined first, and then the similarity characteristic is input into the model, which is beneficial to improving the prediction accuracy of the model.
Optionally, in order to further improve the model recommendation accuracy, the method may further include:
111. and determining the corresponding heat degree of the historical starting and ending points according to the quantity of the plurality of second historical travel records.
Wherein the training sample further comprises: the heat.
Optionally, the method may further include:
112. and according to the portrait data of the trip object, determining a first distance from the departure position carried in the query request to the home of the trip object and a second distance from the departure position to the working unit of the trip object.
The representation data may include representation data for home and work units. The positions of the home and the work units are determined according to the image data of the home and the work units, and then the first distance and the second distance can be determined.
Correspondingly, the training sample further comprises: the first distance and the second distance. In this embodiment, the image data is not directly input to the model, but is preprocessed by determining the first distance and the second distance and inputting the first distance and the second distance to the model, which contributes to improving the prediction accuracy of the model.
In an example, the training samples may also include: the image data. The representation data may also include consumption indices of the travel objects, holiday and non-holiday travel behaviors, destination travel preferences, and the like. Travel object representation data can be mined based on historical driving, historical riding, historical searching and historical driving data of travel objects. Therefore, richer basis information can be provided for the model, and the recommendation accuracy of the model is improved.
Fig. 2 is a flowchart illustrating a destination recommendation method according to another embodiment of the present application. The execution main body of the method can be a client or a server. The client may be hardware integrated on the terminal and having an embedded program, may also be application software installed in the terminal, and may also be tool software embedded in an operating system of the terminal, which is not limited in this embodiment of the present application. The terminal can be any terminal equipment including a mobile phone, a tablet computer, a vehicle-mounted terminal equipment and the like. The server may be a common server, a cloud, a virtual server, or the like, which is not specifically limited in this embodiment of the application. As shown in fig. 2, the method includes:
201. and acquiring a query request of the travel object.
202. And recalling historical starting and ending point pairs of the travel object based on the query request.
203. And determining input information based on the starting characteristics corresponding to the historical starting and ending point pairs.
Wherein the input information includes: and the similarity characteristic is used for representing the similarity between the starting characteristic corresponding to the historical starting and ending point pair and the query request.
204. And inputting the input information into a trained destination recommendation model to obtain a prediction result.
205. And determining whether the terminal point in the historical starting and terminal point pair is a destination to be recommended to the travel object according to the prediction result.
In 201, the query request may be generated when the trip object opens the map software or taxi software on the terminal. The query request may carry object identification information, travel area information, and/or departure information. The departure information may include a departure location, a departure time period, and/or a number of departure weeks, among others. In the present application, the departure information is the current departure information of the trip object, for example: current departure location, current departure time period, and current number of departure weeks.
For the specific implementation of the above-mentioned step 202, reference may be made to corresponding contents in the above-mentioned embodiments, and details are not described herein again.
In the above 203, the determination manner of the similarity characteristic may also be according to the corresponding content in the above embodiments, and is not described herein again.
The specific content included in the input information is the same as the content included in the training sample, and reference may be made to the corresponding content in the above embodiments, which is not described herein again.
In 204, the input information is input into the trained destination recommendation model to obtain a prediction result. The internal structure and the processing flow of the destination recommendation model can be referred to in the prior art, and are not described in detail herein.
In 205, it is determined whether the destination in the history starting and destination pairs is the destination to be recommended to the travel object according to the prediction result.
In the technical scheme provided by the embodiment of the application, the greater the similarity between the starting feature corresponding to the historical starting and ending point pair of the travel object and the query request of the travel object, the greater the probability that the travel object goes to the ending point in the historical starting and ending point pair. Then, the similarity characteristic of the similarity between the starting characteristic corresponding to the historical starting and ending point pair of the travel object and the query request of the travel object is input into the destination recommendation model, so that the recommendation accuracy of the destination recommendation model is improved, and the travel efficiency of the travel object can be improved.
Optionally, the method may further include:
206. and according to the portrait data of the trip object, determining a first distance from the departure position carried in the query request to the home of the trip object and a second distance from the departure position to the working unit of the trip object.
Wherein the input information further comprises: the first distance and the second distance.
The specific implementation manner of the foregoing 206 can refer to corresponding contents in the foregoing embodiments, and is not described herein again.
Optionally, the method may further include:
207. and when the terminal point in the history starting and terminal point pair is determined to be the destination to be recommended to the travel object, recommending the terminal point in the history starting and terminal point pair as the recommended destination to the travel object.
208. And receiving a closing request triggered by the travel object aiming at the recommended destination, and acquiring a recommendation prohibition period.
209. And prohibiting destination recommendation to the travel object within the recommendation prohibition period after the travel object triggers a closing request for the recommended destination.
In the embodiment of the application, a negative feedback mechanism is added, when the travel object is unsatisfied with recommendation, the closing control (such as the first closing control 3 in fig. 3 and the second closing control 6 in fig. 4) is clicked, the travel object can not be recommended within the recommendation forbidding period, and long-term negative experience of wrong recommendation on the travel object is effectively reduced.
Here, it should be noted that: the content of each step in the method provided by the embodiment of the present application, which is not described in detail in the foregoing embodiment, may refer to the corresponding content in the foregoing embodiment, and is not described herein again. In addition, the method provided in the embodiment of the present application may further include, in addition to the above steps, other parts or all of the steps in the above embodiments, and specific reference may be made to corresponding contents in the above embodiments, which is not described herein again.
The following will describe the technical solutions provided in the embodiments of the present application by way of example:
the server can count the historical travel data of each travel object on the platform in advance. The statistical process comprises the following steps:
step 301, according to a third history trip record of a trip object U (where U is a trip object identifier), determining a first time period Stage in which the departure time of the OD pair and the third history trip record is located, and a city code of a positioning city of the trip object U.
And 302, determining at least one second historical travel record of which the starting position is located at the position O, the starting time is located at the Stage and the destination position is located at the position D from the historical travel records of the travel object U.
And 303, fusing the historical departure time of the at least one second historical trip record to obtain fused historical departure time.
And step 304, fusing the historical departure positions of the at least one second historical travel record to obtain a fused historical departure position.
Step 305, determining the corresponding heat degree of the OD according to the record number of the at least one second historical travel record;
step 306, determining the number of the departure weeks of the last trip record in the at least one second historical trip record.
And 307, storing the U, the code, the OD pair, the stage, the historical departure time after fusion, the historical departure position after fusion, the departure week number of the last trip record in at least one second historical trip record and the corresponding heat degree of the OD pair in an associated manner to obtain a statistical record.
Note: in practical application, each trip object corresponds to one or more of the statistical records.
One day, after shopping of the traveling object U in the shopping mall M is finished, the traveling object U wants to take a car to go home. Therefore, the trip object U opens the trip software installed on the mobile phone, and the trip software acquires the trip object identifier U of the trip object U, the current departure position of the trip object U, and the current departure time, and sends an inquiry request to the server, where the inquiry request carries the trip object identifier U, the current departure position, and the current departure time. After receiving the query request, the server executes the following steps:
Wherein, the trip characteristic includes: the system comprises a fused historical departure time, a fused historical departure position, and the number of departure weeks of the last trip record in at least one second historical trip record related to the OD pair.
And step 404, obtaining the portrait data of the trip object U.
The portrait data of the trip object U comprises portrait data of a family and a work unit of the trip object U, a consumption index, trip behavior preference values of holidays and non-holidays and destination trip preference values.
And 405, calculating the current distance between the trip object U and the home and the distance between the trip object U and the working unit according to the current departure position and the image data of the trip object U.
And 406, constructing a feature vector X.
For example: the feature vector X is (position similarity, time similarity, week similarity, distance between the current trip object U and the home, distance between the current trip object U and the working unit, stage, consumption index, travel behavior preference values of holidays and non-holidays, destination travel preference value, fused historical departure time, fused historical departure position, OD corresponding heat degree)
And step 407, determining input information of the destination recommendation model according to the constructed feature vector X.
The end point D in the X and corresponding OD pair may be combined to obtain < X, D >, and the input information is determined according to < X, D >.
Usually, the number of the historical starting and ending point pairs of the traveling object U is multiple, and then, a plurality of feature vectors X are constructed, that is, a plurality of < X, D >. These plural < X, D > constitute a feature set, and the feature set is used as input information.
And step 408, inputting the input information into the destination recommendation model, and sequencing the end points of the plurality of historical start and end point pairs by using the destination recommendation model to obtain a sequencing result.
In step 409, the end point of the first ranking may be used as a destination to be recommended to the travel object.
And step 410, recommending the determined destination needing to be recommended to the travel object.
Another embodiment of the present application provides a human-computer interaction method, as shown in fig. 5, the method includes:
501. and responding to the opening operation of the travel object for the map software, and entering a map interface of the map software.
502. And displaying the recommended taxi taking destination and the ordering control in a correlated manner on the map interface.
503. And responding to the triggering operation of the ordering control, and executing the ordering operation of the recommended taxi taking destination.
The map interface entered in 501 is also the home page of the map software.
In the above 502, for example, as shown in fig. 3, the association display may be: a taxi taking card 1 is displayed on a map interface, and a recommended taxi taking destination and a ordering control 2 are displayed on the taxi taking card 1. Wherein, the recommended taxi taking destination and the ordering control 2 can be displayed on the taxi taking card 1 in parallel.
In addition, estimated taxi taking price, red envelope and the like of the recommended taxi taking destination can be displayed on the taxi taking card 1, and the diversion effect is achieved.
In the above 503, the triggering operation may be a click operation, for example. And executing the ordering operation aiming at the recommended taxi taking destination.
In this embodiment, after the trip object opens the map software and enters the map software, the taxi taking control on the map interface is triggered to trigger the terminal to execute the order taking operation for the recommended destination, so that the operation cost of the trip object is effectively reduced, the order taking time of the trip object is reduced, and the trip efficiency of the trip object is improved.
In addition, fig. 4 shows another display interface, in which a bubbling prompt box 4 is displayed in association with a destination input box 5 displayed on a taxi taking interface; the recommended destination is shown in bubble prompt box 4. In response to the triggering operation of the travel object on the bubble prompt box 4, the ordering operation for the recommended destination is executed.
The method for determining the recommended destination may refer to corresponding contents in the above embodiments, and details are not described herein.
Here, it should be noted that: the content of each step in the method provided by the embodiment of the present application, which is not described in detail in the foregoing embodiment, may refer to the corresponding content in the foregoing embodiment, and is not described herein again. In addition, the method provided in the embodiment of the present application may further include, in addition to the above steps, other parts or all of the steps in the above embodiments, and specific reference may be made to corresponding contents in the above embodiments, which is not described herein again.
Fig. 6 shows a block diagram of a model training apparatus according to another embodiment of the present application. As shown in fig. 6, the apparatus includes:
a generating module 601, configured to generate a query request based on a first historical travel record of a travel object;
a recall module 602, configured to recall a historical starting and ending point pair of the travel object based on the query request;
a building module 603, configured to build, based on the departure features corresponding to the historical starting and ending point pairs, training samples corresponding to the historical starting and ending point pairs, where the training samples include: similarity characteristics used for representing the similarity between the starting characteristics corresponding to the historical starting and ending point pairs and the query request;
a marking module 604, configured to mark, according to a real destination corresponding to the query request and an end point in the history start and end points, a training label of a training sample corresponding to the history start and end point;
the training module 605 is configured to train a preset destination recommendation model according to the marked training samples.
Here, it should be noted that: the model training device provided in the above embodiments may implement the technical solutions described in the above method embodiments, and the specific implementation principle of each module or unit may refer to the corresponding content in the above method embodiments, which is not described herein again.
Fig. 7 is a block diagram illustrating a destination recommendation apparatus according to another embodiment of the present application. As shown in fig. 6, the apparatus includes:
an obtaining module 701, configured to obtain a query request of a trip object;
a recall module 702, configured to recall the historical starting and ending point pairs of the travel object based on the query request;
a first determining module 703, configured to determine, based on the departure characteristics corresponding to the historical starting and ending point pairs, input information, where the input information includes: similarity characteristics used for representing the similarity between the starting characteristics corresponding to the historical starting and ending point pairs and the query request;
an input module 704, configured to input the input information into a trained destination recommendation model to obtain a prediction result;
a second determining module 705, configured to determine, according to the prediction result, whether an end point in the historical start-end point pair is a destination to be recommended to the travel object.
Here, it should be noted that: the destination recommendation device provided in the foregoing embodiments may implement the technical solutions described in the foregoing method embodiments, and the specific implementation principle of each module or unit may refer to the corresponding content in the foregoing method embodiments, which is not described herein again.
Fig. 8 shows a schematic structural diagram of an electronic device according to an embodiment of the present application. As shown in fig. 8, the electronic device includes a memory 1101 and a processor 1102. The memory 1101 may be configured to store other various data to support operations on the electronic device. Examples of such data include instructions for any application or method operating on the electronic device. The memory 1101 may be implemented by any type or combination of volatile or non-volatile memory devices such as Static Random Access Memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic or optical disks.
The memory 1101 is used for storing programs;
the processor 1102 is coupled to the memory 1101, and configured to execute the program stored in the memory 1101, so as to implement the model training method and the destination recommendation method provided in the foregoing method embodiments.
Further, as shown in fig. 8, the electronic device further includes: communication components 1103, display 1104, power components 1105, audio components 1106, and the like. Only some of the components are schematically shown in fig. 8, and the electronic device is not meant to include only the components shown in fig. 8.
Accordingly, the present application further provides a computer-readable storage medium storing a computer program, which when executed by a computer, can implement the steps or functions of the model training method and the destination recommendation method provided in the foregoing method embodiments.
The above-described embodiments of the apparatus are merely illustrative, and the units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of the present embodiment. One of ordinary skill in the art can understand and implement it without inventive effort.
Through the above description of the embodiments, those skilled in the art will clearly understand that each embodiment can be implemented by software plus a necessary general hardware platform, and certainly can also be implemented by hardware. With this understanding in mind, the above-described technical solutions may be embodied in the form of a software product, which can be stored in a computer-readable storage medium such as ROM/RAM, magnetic disk, optical disk, etc., and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the methods described in the embodiments or some parts of the embodiments.
Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present application, and not to limit the same; although the present application has been described in detail with reference to the foregoing embodiments, it should be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions in the embodiments of the present application.
Claims (14)
1. A method of model training, comprising:
generating a query request based on a first historical travel record of a travel object;
recalling historical starting and ending point pairs of the travel object based on the query request;
based on the starting features corresponding to the historical starting and ending point pairs, constructing training samples corresponding to the historical starting and ending point pairs, wherein the training samples comprise: similarity characteristics used for representing the similarity between the starting characteristics corresponding to the historical starting and ending point pairs and the query request;
marking training labels of corresponding training samples of the historical starting and ending points according to the real destination corresponding to the query request and the ending point in the historical starting and ending points;
and training a preset destination recommendation model according to the marked training samples.
2. The method of claim 1, further comprising:
acquiring more than one second historical travel record of the travel object, wherein the historical departure positions of all the second historical travel records are located in the first geographic grid, and the historical destination positions are located in the second geographic grid;
and determining the starting characteristics corresponding to the historical starting and ending point pairs according to the historical starting information of the second historical travel record.
3. The method of claim 2, wherein the historical departure information includes a historical departure location;
determining the departure characteristics corresponding to the historical start and end points according to the historical departure information of the second historical travel record, including:
and fusing the historical starting positions of the second historical travel record to obtain a fused historical starting position which is used as one of the starting characteristics corresponding to the historical starting and ending point pairs.
4. The method of claim 3, wherein the fusing the historical departure positions in the second historical travel record to obtain a fused historical departure position comprises:
determining the radius of a minimum enclosing circle taking the historical departure position as the center of the circle for each historical departure position in the historical departure positions of the second historical travel record; the minimum bounding circle refers to a minimum circle that bounds at least a set number of the historical departure positions; the set number is equal to the product of the number of the second historical travel records and a set preset proportional coefficient;
and determining the center of the minimum enclosing circle with the minimum radius as the fused historical departure position.
5. The method of claim 3, wherein the similarity features include location similarity features characterizing a similarity between the fused historical departure location and the departure location carried in the query request.
6. The method of any of claims 2-5, wherein the historical departure information includes a historical departure time;
determining the departure characteristics corresponding to the historical start and end points according to the historical departure information of the second historical travel record, and further comprising:
and fusing the historical departure time of the second historical travel record to obtain fused historical departure time which is used as one of the departure characteristics corresponding to the historical starting and ending point pairs.
7. The method of claim 6, wherein fusing the historical departure time in the second historical travel record to obtain a fused historical departure time, comprises:
respectively mapping the historical departure time of the second historical travel record into angles under a polar coordinate system;
obtaining a unit vector under a rectangular coordinate system through coordinate conversion according to the angle;
obtaining a target angle under the polar coordinate system through coordinate conversion according to the mean vector of the unit vector;
and mapping the target angle to target time to serve as the historical departure time after fusion.
8. The method of claim 6, wherein the similarity features include a time similarity feature characterizing a similarity between the fused historical departure time and the departure time carried in the query request.
9. The method of any of claims 1-5, wherein the destination recommendation model is a gradient-boosting tree model.
10. The method according to any one of claims 1 to 5, wherein the history starting and ending point pairs are plural;
according to the real destination corresponding to the query request and the end point in the historical start and end points, marking the training label of the training sample corresponding to the historical start and end point, including:
sequencing the plurality of historical starting and ending point pairs according to the distances between the real destination corresponding to the query request and the ending points in the plurality of historical starting and ending points respectively;
and respectively marking the training labels of the training samples corresponding to the plurality of historical starting and ending point pairs according to the sequencing results of the plurality of historical starting and ending point pairs.
11. A destination recommendation method, comprising:
acquiring a query request of a trip object;
recalling historical starting and ending point pairs of the travel object based on the query request;
determining input information based on the departure characteristics corresponding to the historical start and end points, wherein the input information comprises: similarity characteristics used for representing the similarity between the starting characteristics corresponding to the historical starting and ending point pairs and the query request;
inputting the input information into a trained destination recommendation model to obtain a prediction result;
and determining whether the terminal point in the historical starting and terminal point pair is a destination to be recommended to the travel object according to the prediction result.
12. The method of claim 11, further comprising:
determining a first distance from a starting position carried in the query request to a home of the trip object and a second distance from a working unit of the trip object according to the portrait data of the trip object;
the input information further includes: the first distance and the second distance.
13. The method of claim 11, further comprising:
when the terminal point in the history starting and terminal point pair is determined to be the destination to be recommended to the travel object, recommending the terminal point in the history starting and terminal point pair to the travel object as a recommended destination;
receiving a closing request triggered by the travel object aiming at the recommended destination, and acquiring a recommendation prohibition period;
and prohibiting destination recommendation to the travel object within the recommendation prohibition period after the travel object triggers a closing request for the recommended destination.
14. A computer-readable storage medium storing a computer program, wherein the computer program is capable of implementing the model training method of any one of claims 1 to 10 or the destination recommendation method of any one of claims 11 to 13 when executed by a computer.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110844528.7A CN113742608A (en) | 2021-07-26 | 2021-07-26 | Model training method, destination recommendation method, and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110844528.7A CN113742608A (en) | 2021-07-26 | 2021-07-26 | Model training method, destination recommendation method, and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113742608A true CN113742608A (en) | 2021-12-03 |
Family
ID=78729083
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110844528.7A Pending CN113742608A (en) | 2021-07-26 | 2021-07-26 | Model training method, destination recommendation method, and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113742608A (en) |
-
2021
- 2021-07-26 CN CN202110844528.7A patent/CN113742608A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106875066B (en) | Vehicle travel behavior prediction method, device, server and storage medium | |
| WO2020228706A1 (en) | Fence address-based coordinate data processing method and apparatus, and computer device | |
| US10160321B2 (en) | Methods, circuits, devices, systems and associated computer executable code for driver decision support | |
| US11305780B2 (en) | Road condition status prediction method, device, and server, and storage medium | |
| CN107167136B (en) | Position recommendation method and system for electronic map | |
| US20180018572A1 (en) | Method, apparatus, device, and system for predicting future travel volumes of geographic regions based on historical transportation network data | |
| US10119826B2 (en) | System and method for accelerating route search | |
| CN112236787B (en) | System and method for generating personalized destination recommendations | |
| CN111612122B (en) | Real-time demand prediction method and device and electronic equipment | |
| CN110390415A (en) | A kind of method and system carrying out trip mode recommendation based on user's trip big data | |
| CN110222893B (en) | Method and device for recommending delivery places of shared traffic resources and electronic equipment | |
| CN108701413A (en) | Verification picks up the time in the feeding of real-time rideshare | |
| CN109902213B (en) | Real-time bus service line recommendation method and device and electronic equipment | |
| CN109255564A (en) | Pick-up point address recommendation method and device | |
| CN110598917B (en) | Destination prediction method, system and storage medium based on path track | |
| CN111931077B (en) | Data processing method, device, electronic equipment and storage medium | |
| CN109084795A (en) | The method and device of search service facility based on Map Services | |
| US10685364B2 (en) | Method and apparatus for causing a recommendation of a point of interest | |
| WO2021191685A2 (en) | System and method for vehicle event data processing for identifying parking areas | |
| US20230126317A1 (en) | System and method for processing vehicle event data for improved journey trace determination | |
| US20230087199A1 (en) | Method and internet of things system of charging service for new energy vehicle in smart city | |
| CN110674990B (en) | Instant distribution path selection method and system with sliding window updating mechanism | |
| CN112926796B (en) | Get-off point recommendation method and device based on specific scene | |
| CN114707506A (en) | Sorting method and system for express entering delivery site and readable storage medium | |
| Hu et al. | A comprehensive survey of recommendation system based on taxi GPS trajectory |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| TA01 | Transfer of patent application right |
Effective date of registration: 20240311 Address after: # 03-06, Lai Zan Da Building 1, 51 Belarusian Road, Singapore Applicant after: Alibaba Innovation Co. Country or region after: Singapore Address before: Room 01, 45th Floor, AXA Building, 8 Shanton Road, Singapore Applicant before: Alibaba Singapore Holdings Ltd. Country or region before: Singapore |
|
| TA01 | Transfer of patent application right |