CN110287425B

CN110287425B - Peripheral interest point recommendation method for saving calculation resources of vehicle-mounted terminal of internet automobile

Info

Publication number: CN110287425B
Application number: CN201910384508.9A
Authority: CN
Inventors: 张淼; 徐国爱; 秦子彧; 李聪依
Original assignee: Beijing University of Posts and Telecommunications
Current assignee: Beijing University of Posts and Telecommunications
Priority date: 2019-05-09
Filing date: 2019-05-09
Publication date: 2021-05-11
Anticipated expiration: 2039-05-09
Also published as: CN110287425A

Abstract

The present invention relates to a method for recommending surrounding points of interest that saves computing resources of an on-board terminal of a connected car. It combines the possible common categories of travel in daily life, and, under the support of Meituan user data, screens each category of POI to effectively improve word-of-mouth. The ranking of excellent stores in the POI recommendation group achieves the effect of optimizing the quality of recommended POIs; adding sorting data based on public comment data and Tencent map social check-in data, and adding social life-oriented weights to different categories of POI classifications as a new feature. , so as to obtain POI recommendation groups with more comprehensive types and more reasonable sorting. Taking into account the computing power of the on-board terminal, the web API interface is selected for data query function; the detailed information of the recommended location is broadcast to the driver through voice broadcast, which reduces the impact on the driver's attention and driving action, and makes the entire recommendation process. More in line with the safety and entertainment needs of life.

Description

Peripheral interest point recommendation method for saving calculation resources of vehicle-mounted terminal of internet automobile

Technical Field

The invention relates to the field of network communication, in particular to a peripheral interest point recommendation method for saving computing resources of a networked automobile vehicle-mounted terminal.

Background

In the world today, the traditional manufacturing industry gradually upgrades to "smart manufacturing" in the context of the trend of internet multimodal development and industrial intelligence. Under the background, the automobile industry is promoted to develop towards intellectualization and networking under the promotion of technologies such as mobile interconnection, big data and cloud computing. The internet automobile is a new direction of innovation development, brings the automobile industry into a high-speed development period of multi-field and large-system integration, and plays roles in improving traffic efficiency and improving automobile riding experience. At present, the material culture life of Chinese people is greatly improved, the self-driving tour becomes a common leisure mode, and the demand for recommending points of interest (POI) is also improved. At present, a third-party map application can be carried through an intelligent networked automobile vehicle-mounted information system, and POI recommendation is carried out by combining positioning points. At present, mainstream vehicle-mounted third-party map applications comprise a high-resolution map, an Tencent map, a Baidu map and the like, and the function of searching POI based on geographic positions is realized. In the following, a Baidu map is taken as an example to describe how to search for POIs

Inverse geocoding

The Baidu map carries parameters returned by the POI when the Baidu map is set for online reverse geographic query through a web API, and a user can obtain 10 displayed POI by adding a radius parameter which sets a POI parameter value to be 1 and is less than 1000 in a request URL as a radius value of a recalled POI.

Since the interface is not a specific POI query interface, only the recall radius (maximum value is 1000 meters) can be configured as a parameter, and the number of obtained POIs is 10, which is a fixed value.

Another method for obtaining POIs by reverse geocoding is that a geo code new instance () and an on geo code result listener of a reverse geocoding monitoring device of a reverse geocoding retrieval example are firstly created by integrating an SDK of a hundred-degree map in an item, then the reverse geocoding retrieval is carried out by a reverse geocode method, and 10 nearby POIs with the maximum recall radius of 1000 meters are obtained in a callback.

Interface special for POI search

Through web API official development documents of the Baidu map, four different special POI retrieval modes can be known, and the four special POI retrieval modes are respectively served for administrative division area retrieval, circular area retrieval, rectangular area retrieval and location detail retrieval. The corresponding request url format is summarized as follows:

the meaning of each parameter and the meaning of each value in the returned result are described in detail in the Baidu map developer document, and are not described in detail herein. In the four types of interfaces for inquiring the POI, the first three types are to perform the inquiry of the POI result according to the range defined by the geographic coordinates, the location detail retrieval is to perform the retrieval inquiry on the details of the POI according to the UID value of the inquired POI, and generally, the inquiry operation needs to be performed on the basis of obtaining the UID by performing the first three types of inquiry services.

The Baidu map SDK realizes four POI query methods corresponding to the web API (application program interface) POI retrieval method one to one. All four retrieval methods need to set different parameters in poicisysenterrupt on the premise of creating a poiisearch.

After obtaining the search return value, the detailed information to all POIs by the getpoilistmethod in the callback function of the ongetpoiseaarcharesultlistener.

The defects of the prior art are as follows:

1. inverse geocoding method

The Baidu map reverse geocoding interface is not a special POI query interface, the configurable POI filtering and screening parameters only include recall radius (the maximum value is 1000 meters), and the number of the obtained POIs is 10.

The flexibility of the Goodpasts map in reverse geographic query by using a web API is slightly higher than that of a hundred-degree map, the types of the POI can be configured to be returned, and the arrangement sequence of the POI is optimized through a homeorcorp parameter. However, there are also two problems: the fixed value of the return quantity of POI is 30; the arrangement of the homeorcorp parameter can influence the ranking strategy of recalling POI content, and the POI content related to the home is preferentially returned and the POI content related to the company is preferentially returned.

The flexibility of the Tencent map in reverse geographic query by using a web API is highest, and the number of pages and the number of pieces per page are increased by parameters supporting configuration, so that the returned number is not limited; in the ordering policy of returning POI, the update provides a social check-in ordering policy according to social resources thereof, and has high reference significance in the actual geographic interest point recommendation situation; there is still a problem of simplification of the POI return value types of the regions, and there is a limitation, for example, for a region with many office buildings, all POIs of office building types are recommended, and no other types exist.

2. Special POI (Point of interest) query method

The method for carrying out POI inquiry based on the SDK has the advantages that the functions of the methods provided by the map platforms are different. The POI provided by the method has strong category pertinence, but the returned results are fixed to be a single category specified according to the query keyword and are sorted and output according to the distance from the fixed point, no other sorting strategies can be selected, the returned categories of the POI are simplified, and the sequence of the recommended results does not take the social attribute heat of the POI into account.

Through the analysis, the method for obtaining the POI in the reverse geographic query mode can better depict the attributes around the geographic position theoretically, but the actually recommended POI has the two problems of less number of pairs and incomplete type covered by the POI; the direct query of POIs around the fixed point has the defects of return type simplification and sorting according to distance and without actually considering the attraction of the position to tourists.

Disclosure of Invention

At present, the geographic interest point recommendation means of the internet vehicle-mounted information system is few, the recommendation results have the problems of small quantity and incomplete covered categories, and the geographic interest point recommendation with higher practicability cannot be provided for the actual self-driving trip.

The living standard of the material culture of China is gradually improved, and the driving becomes one of the common daily travel modes of people. The invention aims to design a set of geographical position recommendation system friendly to a driver, which is oriented to common driving requirements, reduces the influence on the running speed of a vehicle-mounted terminal to the maximum extent under the precondition of considering real-time performance and the performance of a vehicle-mounted terminal of an internet automobile, saves the computing power of the vehicle-mounted terminal, completes the most valuable POI recommendation with the number specified by a user, broadcasts the POI recommendation to the driver by voice, and recommends more comprehensive geographical interest points to the driver at the minimum cost.

The main objective of the invention is to develop a set of POI recommendation system, and obtain the surrounding valuable location information according to the input location information or the current location information, thereby inquiring the current valuable geographical location information and sending the information to the driver in a voice broadcast mode, and reducing the influence on the driving behavior to the maximum extent.

A peripheral interest point recommendation system for saving internet automobile vehicle-mounted terminal computing resources comprises: the device comprises a query module, a type analysis module, a recommendation module and a voice broadcast module, wherein the query module, the type analysis module, the recommendation module and the voice broadcast module are sequentially connected;

the query module: the system comprises a geographic information acquisition module, a geographic information acquisition module and a display module, wherein the geographic information acquisition module is used for acquiring longitude and latitude of a known target geographic point in a geographic coding mode, and acquiring an official recommended POI group by taking the longitude and latitude as a parameter and using an inverse geographic coding method of a hundred-;

the type analysis module: and calculating the occupied weight according to the category of the POI of the target geographic point, wherein the category of the POI of the target geographic point is divided into a first-level industry classification and a second-level industry classification.

The recommendation module: the POI recommendation system is used for classifying the primary industries in the type analysis module, obtaining a POI set with an expanded recall range through a special POI retrieval interface, and reordering the POI set according to the scoring level of the data of the American group and the public comment platform to obtain a POI recommendation set I. For the secondary industry classification in the type analysis module, configuring a sorting strategy as a social sign-in sorting strategy by using the reverse geocoding function of the Tencent map to obtain a POI recommendation set II as a supplement of the POI recommendation set I; and merging the POI recommendation set II and the POI recommendation set I to obtain a POI recommendation set, and redistributing the number of each type based on the weight of each type of the POI for describing the target geographic point obtained in the type analysis module to obtain a final POI recommendation group.

The voice broadcast module: and the system is used for splicing the final detailed information of the POI recommendation group calculated by the recommendation module according to the language, converting the detailed information of the POI into acceptable language information, forming necessary description of the POI and broadcasting the necessary description to the driver by using a voice conversion function.

On the basis of the scheme, the first-level industry classification of the type analysis module comprises food, hotels, shopping, beauty, leisure and entertainment, sports and fitness and automobile service.

On the basis of the scheme, the secondary industry categories of the type analysis module comprise a Chinese restaurant, a star hotel, a shopping center, a resort, a stadium and automobile sales.

A peripheral interest point recommendation method for saving internet automobile vehicle-mounted terminal computing resources comprises the following steps:

step 1: the longitude and latitude of the target geographic point are obtained through the geographic coding mode of the target geographic point, and the official recommendation POI group of the Baidu map for the target geographic point is obtained through inverse geographic coding by taking the longitude and latitude as a parameter;

step 2: and (3) the official recommended POI group obtained in the step (1) counts the weight occupied by the category calculation of the POI of the target geographic point, the category of the POI of the target geographic point is divided into a primary industry category and a secondary industry category, and different processing based on the category is carried out.

And step 3: and (3) for the first-level industry classification in the step (2), obtaining a POI set with an expanded recall range through a special POI retrieval interface, obtaining social scores of the POI set through public comment and an open query interface of a Mei Tuo, and obtaining a POI recommendation set I by taking the social scores as a high priority ranking policy as a basis for ranking.

Step 4, for the secondary industry classification in the step 2, configuring a sorting strategy as a social sign-in sorting strategy by using the reverse geocoding function of the Tencent map to obtain a POI recommendation set II as a supplement of the POI recommendation set I in the step 3;

and 5: and combining the POI recommendation set II with the POI recommendation set I to obtain a POI recommendation set, and redistributing the number of each type based on the weight of each type of the POI used for describing the target geographic point obtained in the step 2 to obtain a final POI recommendation group.

Step 6: and splicing the final detailed information of the POI recommendation group according to the language, and broadcasting the final detailed information of the POI recommendation group to the driver through a voice conversion technology.

On the basis of the scheme, the first-level industry classification in the step 2 comprises food, hotels, shopping, beauty, leisure and entertainment, sports and fitness and automobile service.

On the basis of the scheme, the secondary industry classification in the step 2 comprises a medium-sized restaurant, a star-sized hotel, a shopping center, a resort, a stadium and automobile sales.

The invention has the beneficial effects that:

1. actual driving scene and daily driving demand

The geographical interest point recommendation method provided by the invention combines common possible travel categories in daily life, and under the support of the US group comment user data, the POIs of catering, shopping and leisure entertainment categories are screened, so that the sequence of shops with excellent public praise in a POI recommendation group is effectively improved, the effect of optimizing the quality of the recommended POI is achieved, and the quality of a search result is guaranteed; the method is characterized in that ranking data based on mass comment data and Tencent map social sign-in data are added, and weights facing social life are added to different categories of POI as new ranking standards, so that a POI recommendation group with more comprehensive types and more reasonable ranking is obtained.

2. Performance of vehicle-mounted terminal is saved and user-friendly broadcasting mode

The geographic point recommendation method based on the invention considers the limited computing capacity of the vehicle-mounted terminal, selects a web API interface as much as possible to perform a data query function for the consideration of the real-time performance and the effect of the method, and has the advantages of rapid request response, short delay time, no time for waiting authentication and the like compared with the method for integrating the SDK in the engineering; the recommended places and the detailed information of the recommended places, such as detailed addresses, telephone numbers, group purchase information and the like, are played to the driver in a voice broadcasting mode, so that the influence on the attention and the driving action of the driver can be reduced to the greatest extent, and the whole recommending process is more suitable for the safety and entertainment requirements of actual life.

The key technical points of the invention content to be protected are as follows:

1. extensible POI recommendation types

The obtained POI types are changed according to user data, and if certain unusual types are fixedly contained in a returned result, the POI types can be independently added in a POI searching process in a type adding mode; the number of POI returns to obtain more can also be configured in the code.

2. POI recommendation mode combining social comment data

The POI recommendation method designed by the invention is based on the reverse geographic query and the return of the POI, and is recommended on the basis of combining the Tencent map, the Mei Tuo and the public comment social data, and the obtained recommendation result takes the social popularity of the POI into consideration, so that the POI recommendation method is more practical.

Drawings

The invention has the following drawings:

FIG. 1 illustrates the components of the POI recommendation system.

Detailed Description

The present invention is described in further detail below with reference to fig. 1.

The specific development process of software development is as follows:

step 1: and obtaining the longitude and latitude of the target geographic point in a geographic coding mode, and carrying out inverse geographic coding by taking the longitude and latitude as a parameter to obtain an official recommended POI group of the Baidu map, wherein the official recommended POI group is used for describing the target geographic point. The Baidu map officially recommended POI groups are 10 in total, the types are not distinguished by default, and the POI groups are sorted according to the distance (distance parameter size) from the target geographic point.

Step 2: and counting the types and the number of all POIs of the official recommended POI group to obtain the types and the weights of the POIs for describing the target geographic point. And classifying the types of the POI describing the target geographic point into a primary industry classification and a secondary industry classification, and performing different processing based on categories on the types of the POI describing the target geographic point. (same as above)

And step 3: for POIs classified in the primary industries such as food, hotels, shopping, beauty, entertainment, sports fitness and automobile service, a way of inquiry of an open platform of a group of beauty assessment platform is used, a ranking strategy in retrieval parameters is configured to be high-grade priority, and the POI points with higher surrounding scores are obtained and serve as a recommended POI set I;

and 4, step 4: for the categories which are not related, performing reverse geocoding query by using an Tencent map interface, using the categories configured in the recall parameters and setting a ranking mode to be 4 social signing scenes, performing priority ranking on hot spots checked in by the user to obtain a recommended POI set II, and performing POI detail query by using id in the POI to obtain detailed information of the POI;

and 5: and combining the recommendation set II and the recommendation set I to obtain a POI recommendation set, and distributing the number of each type in the POI recommendation set based on the weights of the different types obtained in the step 2 to recombine to obtain a final POI recommendation group.

Step 6: and performing language splicing on the detailed information in the final POI recommendation group, converting the basic information of the POI into acceptable language information, for example, converting address ═ Beijing sea lake region West soil city road into address of Beijing sea lake region West soil city road, converting the address into audio broadcasting in a Baidu voice conversion mode, and recommending the POI to a driver.

Those not described in detail in this specification are within the skill of the art.

Claims

1. A peripheral interest point recommendation system for saving the computing resources of the on-board terminal of a connected car is characterized in that, comprising: a query module, a type analysis module, a recommendation module and a voice broadcast module, the query module, the type analysis module, the recommendation module Connect with the voice broadcast module in turn;

The query module is used to obtain the latitude and longitude of the target geographic point by geocoding the known target geographic point, and using the latitude and longitude as a parameter to obtain the official recommended POI group through the inverse geocoding method of Baidu map;

The type analysis module: according to the official recommended POI group obtained by the query module, calculate the weight of the POI category of the target geographic point, and the POI category of the target geographic point is divided into a first-level industry classification and a second-level industry classification;

The recommendation module: used to classify the first-level industries in the type analysis module, obtain the POI collection with the expanded recall range through the special interface for POI retrieval, and reorder the POI collection according to the data scores of Meituan and Dianping Platform. , get the POI recommendation set I; for the secondary industry classification in the type analysis module, by using the inverse geocoding function of Tencent Maps, configure the sorting strategy as the social check-in sorting strategy, and obtain the POI recommendation set II, as a supplement to the POI recommendation set I ; Combine the POI recommendation set II and the recommendation set I to obtain the POI recommendation set, and redistribute the number of each type based on the weight of each type of POI used to describe the target geographic point obtained in the type analysis module to obtain the final POI recommended group;

The voice broadcast module: used to splicing the detailed information of the final POI recommendation group calculated by the recommendation module according to the language, converting the detailed information of the POI into acceptable language information, forming the necessary description of the POI, and using The voice conversion function announces to the driver.

2. The peripheral point-of-interest recommendation system for saving the computing resources of the on-board terminal of a connected car as claimed in claim 1, wherein the first-level industry classification in the type analysis module includes gourmet, hotel, shopping, beauty, leisure and entertainment , sports fitness and car service.

3. The peripheral point of interest recommendation system for saving the computing resources of the on-board terminal of a connected car as claimed in claim 1, wherein the secondary industry classification in the type analysis module includes Chinese restaurants, star-rated hotels, shopping malls, Resorts, stadiums and car sales.

4. A method for recommending surrounding points of interest for saving computing resources of an on-board terminal of a connected car, characterized in that it comprises the following steps:

Step 1: Obtain the latitude and longitude of the target geographic point by geocoding, and use the latitude and longitude as parameters to obtain the official recommendation POI group of Baidu Maps for the target geographic point through reverse geocoding;

Step 2: For the officially recommended POI group obtained in Step 1, calculate the weight of the POI category of the target geographic point. The POI category of the target geographic point is divided into first-level industry classification and second-level industry classification, and different processing based on the category is performed. ;

Step 3: For the first-level industry classification in Step 2, obtain the POI set with the expanded recall range through the special interface for POI retrieval, and obtain the social score of the POI set through the public comment and the open query interface of Meituan, and the ranking strategy is high score. Priority, and use this as the basis for reordering to obtain POI recommendation set I;

Step 4: For the secondary industry classification in step 2, by using the reverse geocoding function of Tencent Maps, configure the sorting strategy as a social check-in sorting strategy, and obtain POI recommendation set II as a supplement to POI recommendation set I in step 3;

Step 5: Combine the POI recommendation set II and the POI recommendation set I to obtain the POI recommendation set. Based on the weights of each type of POI used to describe the target geographic point obtained in step 2, redistribute the number of each type to obtain the final The POI recommendation group;

Step 6: The detailed information of the final POI recommendation group is spliced according to the language, and the detailed information of the final POI recommendation group is broadcast to the driver through the voice conversion technology.

5. The method for recommending peripheral points of interest for saving computing resources of the on-board terminal of a connected car as claimed in claim 4, wherein the first-level industry classification in the step 2 includes food, hotel, shopping, beauty, leisure and entertainment, Sports fitness and car service.

6. The method for recommending peripheral points of interest for saving computing resources of the on-board terminal of a connected car as claimed in claim 4, wherein the secondary industry classification in the step 2 includes Chinese restaurants, star-rated hotels, shopping malls, and resorts , sports venues and car sales.