CN111581505B - Flight recommendation method and system based on combined recommendation - Google Patents
Flight recommendation method and system based on combined recommendation Download PDFInfo
- Publication number
- CN111581505B CN111581505B CN202010349887.0A CN202010349887A CN111581505B CN 111581505 B CN111581505 B CN 111581505B CN 202010349887 A CN202010349887 A CN 202010349887A CN 111581505 B CN111581505 B CN 111581505B
- Authority
- CN
- China
- Prior art keywords
- recommendation
- flight
- features
- algorithm
- user
- 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.)
- Active
Links
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
- G06F16/9535—Search customisation based on user profiles and personalisation
-
- 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
- G06F16/9536—Search customisation based on social or collaborative filtering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/06—Buying, selling or leasing transactions
- G06Q30/0601—Electronic shopping [e-shopping]
- G06Q30/0631—Item recommendations
Abstract
The invention discloses a flight recommendation method and system based on combined recommendation, and relates to the field of aviation informatization management. The method comprises the following steps: the client sends the user information and the flight request instruction to the server; the server retrieves the alternative flights meeting the flight demand information from the database, determines at least two recommendation algorithms to be used according to the number of the explicit features and the number of the implicit features contained in the user information, selects corresponding recommendation flights from the alternative flights according to each recommendation algorithm respectively, obtains recommendation results of each recommendation algorithm, combines all obtained recommendation results, and determines the final recommendation flights. The invention can adapt to users with different feature integrality, thereby improving the accuracy and reliability of recommendation, recommending flights which best meet the requirements of the users, saving the time when a target user picks the flights, enabling the recommended flights to meet the requirements of the users, and enabling the recommendation accuracy to be higher and the effect to be better.
Description
Technical Field
The invention relates to the field of aviation informatization management, in particular to a flight recommendation method and system based on combined recommendation.
Background
With the rapid development of passenger aviation, the aircraft gradually becomes the choice of many people when going out, at present, users order at third party websites when ordering flights, the websites integrate the flight resources of a plurality of airlines, then all the flights meeting the conditions are displayed to the users by setting a plurality of simple options, and as the airlines compete fiercely, the number of the flights is quite large, therefore, the data volume faced by the clients when selecting the flights at the websites is quite large, the flights meeting the demands of the clients are difficult to select in a short time, and the time and effort are wasted.
The present flight recommendation system sorts the flights according to a certain screening condition after the target user performs the flight inquiry, the screening condition is usually selected manually by the target user and is limited by the processing capability of the server, and only single condition screening is usually supported, so that the inconvenience is very high, for example, after the target user selects to sort according to the price, whether each flight meets other requirements of the user or not is further judged one by one, if so, whether to provide dining, take-off time, landing time, direct flight or not and the like, the target user wastes a long time, and the most suitable flights are difficult to select from a plurality of flights.
Disclosure of Invention
Aiming at the problems that in the prior art, a user wastes time and labor in the process of selecting flights and is difficult to select flights meeting own needs, the invention provides a flight recommendation method and a flight recommendation system based on combined recommendation, which are used for determining a recommendation algorithm according to the number of dominant features and the number of implicit features contained in user information so as to recommend flights meeting the needs of the user, thereby saving the time when the user selects flights, and enabling the recommended flights to meet various needs of the user.
The technical scheme for solving the technical problems is as follows:
a flight recommendation method based on combined recommendation, comprising:
the client acquires flight demand information of a target user, generates a flight request instruction according to the flight demand information, acquires user information of the target user, and sends the user information and the flight request instruction to a server;
the server receives and analyzes the flight request instruction, invokes the alternative flights meeting the flight demand information from a database, determines at least two recommendation algorithms to be used according to the number of explicit features and the number of implicit features contained in the user information, selects corresponding recommendation flights from the alternative flights according to each recommendation algorithm respectively, obtains recommendation results of each recommendation algorithm, combines all obtained recommendation results, determines a final recommendation flight, and sends the final recommendation flight to the client;
the client receives and displays the final recommended flight.
The other technical scheme for solving the technical problems is as follows:
a combined recommendation based flight recommendation system comprising: client, server and database, wherein:
the client is used for acquiring flight demand information of a target user, generating a flight request instruction according to the flight demand information, acquiring user information of the target user, and sending the user information and the flight request instruction to a server;
the server is used for receiving and analyzing the flight request instruction, retrieving the alternative flights meeting the flight demand information from the database, determining at least two recommendation algorithms to be used according to the number of the explicit features and the number of the implicit features contained in the user information, selecting the corresponding recommendation flights from the alternative flights according to each recommendation algorithm respectively to obtain the recommendation result of each recommendation algorithm, combining all the obtained recommendation results, determining the final recommendation flights, and sending the final recommendation flights to the client;
the client is also used for receiving and displaying the final recommended flight.
The beneficial effects of the invention are as follows: according to the flight recommendation method and system based on combined recommendation, the recommendation algorithm is determined according to the number of the explicit features and the number of the implicit features contained in the user information, so that the recommendation algorithm which is most suitable for the current user can be determined according to the completeness of the user information.
Additional aspects of the invention will be set forth in part in the description which follows and, in part, will be obvious from the description, or may be learned by practice of the invention.
Drawings
FIG. 1 is a flow chart of a flight recommendation method based on combined recommendation according to an embodiment of the present invention;
FIG. 2 is a schematic diagram of a similar user group formation provided by an embodiment of a flight recommendation method based on combined recommendation according to the present invention;
FIG. 3 is a schematic diagram of a structural framework provided by an embodiment of a flight recommendation system based on combined recommendations in accordance with the present invention;
fig. 4 is a schematic diagram of network topology connection provided by an embodiment of a flight recommendation system based on combined recommendation according to the present invention.
Detailed Description
The principles and features of the present invention are described below with reference to the drawings, the illustrated embodiments are provided for illustration only and are not intended to limit the scope of the present invention.
As shown in fig. 4, an exemplary network topology connection schematic is provided, and the server 10 of the present invention, which is based on the combined recommendation flight recommendation method, may be connected to a plurality of clients 20 through the internet, and an exemplary communication link establishment method is provided below.
The client 20 first sends an access internet request to the internet access gateway 30, and the internet access gateway 30 sends the access request to the internet service gateway 40, and the internet service gateway 40 authenticates the access request. After the authentication is successful, a message of the authentication success is sent to the internet access gateway 30, and a network communication transmission channel is established with the internet access gateway 30. The internet access gateway 30 transmits a message of successful authentication to the client 20, and the client 20 establishes a communication link with the server 10 after receiving the message of successful authentication, and performs data transmission.
Some possible embodiments of the invention are described below based on the above network structure.
As shown in fig. 1, a flow chart provided for an embodiment of a flight recommendation method based on combined recommendation according to the present invention is implemented based on combined recommendation, and includes:
s1, a client acquires flight demand information of a target user, generates a flight request instruction according to the flight demand information, acquires user information of the target user, and sends the user information and the flight request instruction to a server.
The flight demand information may be information capable of identifying flights, for example, may be a departure place and a destination, and if the destination user inputs the departure place a and the destination B through the client, and takes the departure place a and the destination B as the flight demand information, all flights with the departure place a and the destination B may be identified.
It should be appreciated that the user information may include personal information and usage information related to the target user, such as age, occupation, birthday, background, page stay time, mouse click region, flight view status, collection status, etc. of the target user.
Specifically, a pre-written program plug-in can be built in the client, and when a target user uses software, APP or a webpage to operate, an operation instruction and a movement track of a mouse can be captured through the plug-in, so that the page residence time, the mouse clicking area, the flight viewing condition, the collection condition and the like of the target user are determined according to page structures and contents.
The age, occupation, birthday, background, etc. of the target user may be manually entered by the target user.
S2, the server receives and analyzes the flight request instruction, and the alternative flights meeting the flight demand information are called from the database.
After the flight request instruction is analyzed, the server restores the flight demand information, and detects the corresponding flight from the first database as an alternative flight by taking the flight demand information as a search condition.
For example, assuming that the target user's flight demand is x months x days from Shanghai to Beijing, then the time may be: x month x day, place of departure: shanghai, arrival at the ground: and searching Beijing by taking the three values as search conditions, and selecting flights with the total time of x months and x days, the departure place of Shanghai and the arrival place of Beijing from the first database as alternative flights.
And S3, the server determines at least two recommendation algorithms to be used according to the number of the explicit features and the number of the implicit features contained in the user information.
It should be appreciated that the explicit features may be age, occupation, birthday, background, etc. of the target user, and the implicit features may be page stay time, mouse click area, flight viewing status, collection status, etc. of the target user.
The explicit features can be input by a target user, and the implicit features can be obtained by monitoring through the plug-in, which are described earlier and are not repeated.
Taking the explicit feature as an example, there may be some users whose personal information is not complete, and thus this difference in number may be present. For users with more dominant features, the recommendation can be easier, algorithms such as collaborative filtering recommendation, history purchase record recommendation, social network recommendation and the like can be selected for direct recommendation, and for users with less dominant information, recommendation algorithms such as knowledge recommendation and content recommendation can be used for combined recommendation according to the user information detailed degree of the users, more reasonable and proper recommendation algorithms can be used for combined recommendation, the success rate of recommendation is further improved, and the actual demands of the users are met.
For example, the number of dominant features exceeds a threshold, the number of recessive features does not exceed a threshold, and a+b algorithm combination may be recommended; the number of dominant features exceeds a threshold, the number of recessive features also exceeds a threshold, a+c algorithm combination may be recommended, and so on.
It should be understood that what kind of recommendation algorithm is selected specifically may be selected to be combined according to the actual setting, if the dominant feature is too small, then the algorithm that recommends more depending on the recessive feature may be selected to be combined, if the recessive feature is too small, then the algorithm that recommends more depending on the dominant feature may be selected to be combined, and the optional recommendation algorithm may include: content-based recommendations, collaborative filtering-based recommendations, rule-based recommendations, knowledge-based recommendations, social network-based recommendations, or historic purchase record-based recommendations, etc.
These recommendation algorithms are briefly described below.
The theoretical basis of the content-based recommendation method mainly comes from information retrieval and information filtering, and the content-based recommendation method recommends recommended items to the user, which are not touched by the user, according to the browsing records of the user in the past. Wherein the implicit features can reflect the user's browsing records to some extent and thus can be used for content recommendation. For example, it may be calculated using TF-IDF (term frequency-inverse text frequency index) method, the characteristic attribute of the user is obtained according to the access record of the user, then the value of the utility function is calculated according to the characteristic attribute of the user and the characteristic attribute of the recommended item, and the result is recommended to the user.
Based on the many types of collaborative filtering recommendations, the following description will simply be made based on collaborative filtering by users, the basic idea is that if some users score some items consistently or closely, they can be considered to have a smaller difference in scoring other items, and further, the scoring values of these similar users' items can be used to estimate the non-scored items of the target user. The nearest neighbor user set with similar interest preference to the target user is searched by using a mathematical statistics method based on collaborative filtering of the users, then the score of the target user to the specific item is predicted by using a certain mathematical method based on the score of the nearest neighbor user to the specific item, and the top N commodities with the highest predicted score can be regarded as top-N commodities which are most likely to be interested by the user and returned to the target user.
Specifically, firstly modeling according to the scoring condition of the users on the items, effectively measuring the similarity among the users, then searching a nearest neighbor user set of the target user, finally calculating the scoring of the target user on the unscored items according to the scoring information of the nearest neighbor user of the target user on the items through a corresponding mathematical formula, and recommending N items with the highest scoring.
As shown in fig. 2, a schematic diagram of forming a similar user group is provided, the similarity between the target user and other target users is determined, the target user represented by a white circle is far away from the target user, the similarity is low, the target user represented by a black circle is close to the target user, the similarity is high, and therefore the similar user group can be determined through the formula, namely, the black small circles in the large circle form the similar user group, and each black small circle is a target user similar to the target user in the similar user group.
For example, the nearest neighbor user set may be determined by the following formula:
firstly, a set of all items scored by the target users i and j is obtained, and then the similarity degree between the target users i and j is measured through different similarity calculation methods and is marked as sim (i, j).
An exemplary method of computing sim (i, j) is given below:
set I ij Representing the set of items that are scored together by user i and user j, the similarity sim (i, j) between user i and user j can be derived using the Pearson correlation coefficient calculation formula:
wherein, the liquid crystal display device comprises a liquid crystal display device,
representing the average score, r, for user i and user j, respectively ic Representing the score of user i for item c, user j for item c is scored by r jc And (3) representing.
For another example, the predictive score of the target user for a preset item may be determined by:
assume that the nearest neighbor set of target item I uses i= { I 1 ,i 2 ,...,i k Expressed by the term "user" then the predictive score p for the target user u for item i can be calculated from the scores of user u for items in the nearest neighbor set ui The specific method comprises the following steps:
where sim (i, k) represents the similarity between the target item i and the nearest neighbor item k, r uk Representing the score of user u for item k,
representing the average scores of all users for item i and item k, respectively.
Based on knowledge recommendation, an expert knowledge base is added in a traditional recommendation algorithm, and the method is suitable for recommendation in specific fields, such as the field of flight recommendation, and because the expert knowledge base is introduced, the degree of dependence on explicit features and implicit features of a user is low, the accuracy of recommendation is high, and commodity features can be subjected to semantic expansion by introducing the expert knowledge base, and content with similar semantics can be recommended.
Based on the fact that social network recommendation is a recommendation mode which is newly developed in recent years, dependence on explicit characteristics of users is slightly high, the principle is that social network information of the users is added into a traditional recommendation algorithm, and then articles interested by the friends are recommended to the users according to similarity among the friends in the social network of the users.
The rest recommended algorithm is not more involved in the invention and is not repeated.
S4, the server selects the corresponding recommended flight from the candidate flights according to each recommended algorithm respectively, and a recommended result of each recommended algorithm is obtained.
S5, the server combines all the obtained recommended results, determines the final recommended flight, and sends the final recommended flight to the client.
For example, assuming that the flights selected by the device are A, B, C, D and E, and a combination of two algorithms of content-based recommendation and knowledge-based recommendation is used, then a recommended flight may be selected from the candidate flights by using the content-based recommendation algorithm, and assuming that the flights recommended by the content-based recommendation algorithm are flights A, B and C and the flights recommended by the knowledge-based recommendation algorithm are flights C, D and E, several alternative combinations are given below.
In the first mode, the number of occurrences of each flight in the recommendation result obtained by each recommendation algorithm can be counted, and a plurality of flights with the largest number of occurrences are used as final recommended flights.
For example, flight C appears 2 times, both based on the content recommendation algorithm and based on the knowledge recommendation algorithm, and the remaining flights appear only once, then flight C may be considered as the final recommended flight.
In the second mode, a preset number of flights may be selected as final recommended flights, for example, assuming that 3 recommended flights are selected as final recommended flights, then, since the number of occurrences of the flight C is the largest, it is difficult to take the flight C as one of the final recommended flights, and the remaining 2 final recommended flights may be randomly generated on average from the two recommended results, for example, assuming that the recommended flights randomly generated from the recommended results are the flight a and the flight D, respectively, then the flight a, the flight D, and the flight C may be the most final recommended flights.
And S6, the client receives and displays the final recommended flight.
Compared with the traditional flight recommendation method, the flight recommendation method provided by the embodiment of the invention can adapt to users with different feature integrality according to different persons, thereby improving the recommendation accuracy, improving the recommendation reliability through the combined recommendation of a plurality of recommendation algorithms, recommending flights which are most in line with the requirements of the users, saving the time when a target user picks flights, enabling the recommended flights to meet the requirements of the users, and not needing to analyze the flight selection preference of each user independently, thereby saving system resources, ensuring higher recommendation accuracy and better effect.
Optionally, in some possible embodiments, the combination of all the obtained recommended results, and determining the final recommended flight specifically includes:
the server counts the occurrence times of each recommended flight in all the recommended results, and takes the top n recommended flights with the highest occurrence times as final recommended flights, wherein n is more than or equal to 1.
It should be understood that the number of n may be set according to actual requirements.
For example, if the flights are A, B, C, D and E, and a combination of three algorithms including content-based recommendation, knowledge-based recommendation and collaborative filtering recommendation is used, the recommended flights may be picked up by the three recommendation algorithms, so as to obtain three recommendation results, and if the flights recommended by the content recommendation algorithm are A, B and C, the flights recommended by the knowledge recommendation algorithm are C, D and E, and the flights recommended by the collaborative filtering recommendation algorithm are A, C and E.
Then assuming that the number of n is 1, i.e. only 1 final recommended flight is recommended to the user, then after statistics, it is found that only flight C is recommended 3 times, and then flight C can be taken as the final recommended flight.
Assuming that the number of n is 3, i.e., 3 final recommended flights are recommended to the user, then by statistics, it is found that flight a is recommended 2 times, flight B is recommended 1 time, flight C is recommended 3 times, flight D is recommended 1 time, flight E is recommended 2 times, and the number of recommended flights A, C and E is the largest, then flights A, C and E may be considered as final recommended flights.
The flight most suitable for the expectations of the user can be obtained by combining the results of the recommendation algorithms, the recommendation accuracy is provided, the recommendation process is simple, only statistics are involved, the occupation of system resources is small, the recommendation speed is high, the instantaneity is high, and the user experience is good.
Optionally, in some possible embodiments, determining at least two recommendation algorithms to use according to the number of explicit features and the number of implicit features contained in the user information specifically includes:
the server judges the number of the explicit features and the number of the implicit features contained in the user information;
when the number of the dominant features does not exceed a first preset threshold value and the number of the recessive features does not exceed a second preset threshold value, the dominant features and the recessive features are complemented by a preset complement algorithm;
when the number of the explicit features does not exceed a first preset threshold value, but the number of the implicit features exceeds a second preset threshold value, determining that the recommendation algorithm is a knowledge-based recommendation algorithm and a content-based recommendation algorithm;
when the number of the explicit features exceeds a first preset threshold value, but the number of the implicit features does not exceed a second preset threshold value, determining that the recommendation algorithm is based on a collaborative filtering recommendation algorithm and a social network recommendation algorithm;
when the number of the explicit features exceeds a first preset threshold and the number of the implicit features exceeds a preset second preset threshold, determining that the recommendation algorithm is at least two of a knowledge-based recommendation algorithm, a content-based recommendation algorithm, a collaborative filtering recommendation algorithm and a social network recommendation algorithm.
It should be appreciated that knowledge-based recommendation algorithms and content-based recommendation algorithms have a weak dependence on the explicit features of the user, and thus, when the explicit features of the user are significantly insufficient, the user may be recommended by the knowledge-based recommendation algorithm and the content-based recommendation algorithm. The dependency of the collaborative filtering recommendation algorithm and the social network recommendation algorithm on the implicit characteristics of the user is weak, so that when the implicit characteristics of the user are obviously insufficient, the user can be recommended through the collaborative filtering recommendation algorithm and the social network recommendation algorithm.
The recommendation method has the advantages that the characteristics of the user are divided into explicit characteristics such as age, occupation, birthday or background and implicit characteristics such as page stay time, mouse click area, flight viewing condition or collection condition, and corresponding recommendation algorithms are selected according to the conditions of the explicit characteristics and the implicit characteristics of the user, so that recommendation of different users can be applicable, recommendation by adopting one algorithm is avoided, when specific information of the user is lost, the problem of reduced recommendation precision is solved, and the practicability is higher.
Optionally, in some possible embodiments, the completing the dominant feature by a preset completion algorithm specifically includes:
the server determines similar user groups of the target user according to the user information;
the server determines the transfer relation among all users in the similar user group through a preset transfer algorithm, selects the similar user with the highest similarity with the target user from the similar user group according to the implicit characteristics of the target user, and complements the explicit characteristics of the target user according to the explicit characteristics of the similar user and the transfer relation.
It should be noted that, by analyzing the association relationship between users, the relationship between users without direct association can be obtained through the transmission principle, but because the target users are not in the similar user group, the target users closest to the target users can be selected through implicit features such as page stay time, mouse click area, flight view condition, collection condition and the like, the operation habit and purchasing habit of the target users are closest to each other, and therefore the association degree between the target users is relatively larger, therefore, the missing dominant features of the target users can be recursively deduced through the transmission relationship between the similar users and other users, so that the missing information of the target users can be accurately complemented, the similar users are matched through the implicit features, the dominant features are further thrust back, the association between the dominant features and the implicit features is fully mined through the similar relationship between the users, and the accuracy of information complement is higher.
Optionally, in some possible embodiment manners, the complement of the implicit feature by a preset complement algorithm specifically includes:
the server determines similar user groups of the target user according to the user information;
the server fills the implicit features missing from the target user by the average value of the corresponding implicit features of the similar user group.
It should be understood that the similar user group is a plurality of users consistent with the current user in purchasing behavior or scoring behavior, if scores of some users on partial projects tend to be consistent or close, the differences of scores of some users on other projects can be considered to be smaller, so that the implicit characteristics of the user missing can be complemented in an average value filling way.
It should be understood that in some possible embodiments, a combination of any or all of the implementations of the above embodiments may be included.
As shown in fig. 3, a schematic structural framework provided for an embodiment of a flight recommendation system based on combined recommendation according to the present invention, the flight recommendation system based on combined recommendation implementation includes: client 1, server 2 and database 3, wherein:
the client 1 is used for acquiring flight demand information of a target user, generating a flight request instruction according to the flight demand information, acquiring user information of the target user, and sending the user information and the flight request instruction to the server 2;
the server 2 is configured to receive and parse a flight request instruction, retrieve an alternative flight meeting the flight demand information from the database 3, determine at least two recommendation algorithms to be used according to the number of explicit features and the number of implicit features contained in the user information, pick out a corresponding recommendation flight from the alternative flights according to each recommendation algorithm, obtain a recommendation result of each recommendation algorithm, combine all obtained recommendation results, determine a final recommendation flight, and send the final recommendation flight to the client 1;
the client 1 is also adapted to receive and display the final recommended flight.
According to the flight recommendation system based on combined recommendation, the recommendation algorithm is determined according to the number of the explicit features and the number of the implicit features contained in the user information, so that the recommendation algorithm which is most suitable for the current user can be determined according to the integrity degree of the user information.
Optionally, in some possible embodiments, the server 2 is specifically configured to count the number of occurrences of each recommended flight in all recommended results, and take the first n recommended flights with the largest number of occurrences as the final recommended flights, where n is greater than or equal to 1.
Optionally, in some possible embodiments, the server 2 is specifically configured to determine the number of explicit features and the number of implicit features contained in the user information;
when the number of the dominant features does not exceed a first preset threshold value and the number of the recessive features does not exceed a second preset threshold value, the dominant features and the recessive features are complemented by a preset complement algorithm;
when the number of the explicit features does not exceed a first preset threshold value, but the number of the implicit features exceeds a second preset threshold value, determining that the recommendation algorithm is a knowledge-based recommendation algorithm and a content-based recommendation algorithm;
when the number of the explicit features exceeds a first preset threshold value, but the number of the implicit features does not exceed a second preset threshold value, determining that the recommendation algorithm is based on a collaborative filtering recommendation algorithm and a social network recommendation algorithm;
when the number of the explicit features exceeds a first preset threshold and the number of the implicit features exceeds a preset second preset threshold, determining that the recommendation algorithm is at least two of a knowledge-based recommendation algorithm, a content-based recommendation algorithm, a collaborative filtering recommendation algorithm and a social network recommendation algorithm.
Optionally, in some possible embodiments, the server 2 is specifically configured to determine a similar user group of the target user according to the user information, determine a transfer relationship between the users in the similar user group through a preset transfer algorithm, and selecting the similar user with the highest similarity with the target user from the similar user group according to the implicit characteristics of the target user, and complementing the explicit characteristics of the target user which are missing according to the explicit characteristics and the transfer relation of the similar user.
Optionally, in some possible embodiments, the server 2 is specifically configured to determine a similar user group of the target user according to the user information, and fill in the implicit features missing from the target user by an average value of the corresponding implicit features of the similar user group.
It should be understood that in some possible embodiments, a combination of any or all of the implementations of the above embodiments may be included.
The foregoing embodiments are product embodiments corresponding to the previous method embodiments, so the description of each optional implementation manner of the product embodiments and the corresponding technical effects of each optional implementation manner of the product embodiments may refer to the previous method embodiments, and are not repeated herein.
The reader will appreciate that in the description of this specification, a description of terms "one embodiment," "some embodiments," "examples," "specific examples," or "some examples," etc., means that a particular feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the invention. In this specification, schematic representations of the above terms are not necessarily directed to the same embodiment or example. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples. Furthermore, the different embodiments or examples described in this specification and the features of the different embodiments or examples may be combined and combined by those skilled in the art without contradiction.
In the several embodiments provided in this application, it should be understood that the disclosed apparatus and method may be implemented in other ways. For example, the method embodiments described above are merely illustrative, e.g., the division of steps is merely a logical function division, and there may be additional divisions of actual implementation, e.g., multiple steps may be combined or integrated into another step, or some features may be omitted or not performed.
The present invention is not limited to the above embodiments, and various equivalent modifications and substitutions can be easily made by those skilled in the art within the technical scope of the present invention, and these modifications and substitutions are intended to be included in the scope of the present invention. Therefore, the protection scope of the invention is subject to the protection scope of the claims.
Claims (8)
1. A flight recommendation method based on combined recommendation, comprising:
the client acquires flight demand information of a target user, generates a flight request instruction according to the flight demand information, acquires user information of the target user, and sends the user information and the flight request instruction to a server;
the server receives and analyzes the flight request instruction, invokes the alternative flights meeting the flight demand information from a database, determines at least two recommendation algorithms to be used according to the number of explicit features and the number of implicit features contained in the user information, selects corresponding recommendation flights from the alternative flights according to each recommendation algorithm respectively, obtains recommendation results of each recommendation algorithm, combines all obtained recommendation results, determines a final recommendation flight, and sends the final recommendation flight to the client;
the client receives and displays the final recommended flight;
the method specifically comprises the steps of determining at least two recommendation algorithms to be used according to the number of dominant features and the number of recessive features contained in the user information, wherein the recommendation algorithms specifically comprise:
the server judges the number of the explicit features and the number of the implicit features contained in the user information;
when the number of the dominant features does not exceed a first preset threshold value and the number of the recessive features does not exceed a second preset threshold value, complementing the dominant features and the recessive features through a preset complementing algorithm;
when the number of the explicit features does not exceed a first preset threshold value, but the number of the implicit features exceeds a preset second preset threshold value, determining that the recommendation algorithm is a knowledge-based recommendation algorithm and a content-based recommendation algorithm;
when the number of the explicit features exceeds a first preset threshold value, but the number of the implicit features does not exceed a second preset threshold value, determining that the recommendation algorithm is based on a collaborative filtering recommendation algorithm and a social network recommendation algorithm;
and when the number of the explicit features exceeds a first preset threshold and the number of the implicit features exceeds a preset second preset threshold, determining that the recommendation algorithm is at least two of a knowledge-based recommendation algorithm, a content-based recommendation algorithm, a collaborative filtering recommendation algorithm and a social network recommendation algorithm.
2. The method for recommending flights based on combined recommendation according to claim 1, wherein the step of combining all the obtained recommended results to determine the final recommended flights comprises the steps of:
the server counts the occurrence times of each recommended flight in all recommended results, and takes the first n recommended flights with the largest occurrence times as final recommended flights, wherein n is more than or equal to 1.
3. The flight recommendation method based on combined recommendation according to claim 2, wherein the explicit feature is complemented by a preset complement algorithm, specifically comprising:
the server determines similar user groups of the target user according to the user information;
and the server determines the transfer relation among all the users in the similar user group through a preset transfer algorithm, selects the similar user with the highest similarity with the target user from the similar user group according to the implicit characteristics of the target user, and complements the explicit characteristics of the target user according to the explicit characteristics of the similar user and the transfer relation.
4. The flight recommendation method based on combined recommendation according to claim 2, wherein the implicit features are complemented by a preset complement algorithm, specifically comprising:
the server determines similar user groups of the target user according to the user information;
and the server fills the missing implicit features of the target user through the average value of the corresponding implicit features of the similar user groups.
5. A flight recommendation system based on combined recommendation, comprising: client, server and database, wherein:
the client is used for acquiring flight demand information of a target user, generating a flight request instruction according to the flight demand information, acquiring user information of the target user, and sending the user information and the flight request instruction to a server;
the server is used for receiving and analyzing the flight request instruction, retrieving the alternative flights meeting the flight demand information from the database, determining at least two recommendation algorithms to be used according to the number of the explicit features and the number of the implicit features contained in the user information, selecting the corresponding recommendation flights from the alternative flights according to each recommendation algorithm respectively to obtain the recommendation result of each recommendation algorithm, combining all the obtained recommendation results, determining the final recommendation flights, and sending the final recommendation flights to the client;
the client is also used for receiving and displaying the final recommended flight;
the server is specifically configured to determine the number of explicit features and the number of implicit features included in the user information;
when the number of the dominant features does not exceed a first preset threshold value and the number of the recessive features does not exceed a second preset threshold value, complementing the dominant features and the recessive features through a preset complementing algorithm;
when the number of the explicit features does not exceed a first preset threshold value, but the number of the implicit features exceeds a preset second preset threshold value, determining that the recommendation algorithm is a knowledge-based recommendation algorithm and a content-based recommendation algorithm;
when the number of the explicit features exceeds a first preset threshold value, but the number of the implicit features does not exceed a second preset threshold value, determining that the recommendation algorithm is based on a collaborative filtering recommendation algorithm and a social network recommendation algorithm;
and when the number of the explicit features exceeds a first preset threshold and the number of the implicit features exceeds a preset second preset threshold, determining that the recommendation algorithm is at least two of a knowledge-based recommendation algorithm, a content-based recommendation algorithm, a collaborative filtering recommendation algorithm and a social network recommendation algorithm.
6. The combined recommendation-based flight recommendation system according to claim 5, wherein the server is specifically configured to count the occurrence number of each recommended flight in all recommendation results, and take the first n recommended flights with the largest occurrence number as final recommended flights, wherein n is greater than or equal to 1.
7. The flight recommendation system based on combined recommendation of claim 6, wherein the server is specifically configured to determine a similar user group of the target user according to the user information, determine a transfer relationship between each user in the similar user group through a preset transfer algorithm, select, from the similar user group, a similar user with highest similarity to the target user according to implicit features of the target user, and complement explicit features of the similar user and the explicit features of the missing target user according to the explicit features of the similar user and the transfer relationship.
8. The combined recommendation-based flight recommendation system according to claim 7, wherein the server is specifically configured to determine a similar user group of the target user according to the user information, and fill in implicit features missing from the target user by an average value of corresponding implicit features of the similar user group.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010349887.0A CN111581505B (en) | 2020-04-28 | 2020-04-28 | Flight recommendation method and system based on combined recommendation |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010349887.0A CN111581505B (en) | 2020-04-28 | 2020-04-28 | Flight recommendation method and system based on combined recommendation |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111581505A CN111581505A (en) | 2020-08-25 |
| CN111581505B true CN111581505B (en) | 2023-07-07 |
Family
ID=72116780
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010349887.0A Active CN111581505B (en) | 2020-04-28 | 2020-04-28 | Flight recommendation method and system based on combined recommendation |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111581505B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112200625A (en) * | 2020-09-30 | 2021-01-08 | 中国民航信息网络股份有限公司 | Flight resource recommendation method and device |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110569447A (en) * | 2019-09-12 | 2019-12-13 | 腾讯音乐娱乐科技（深圳）有限公司 | network resource recommendation method and device and storage medium |
| CN110866181A (en) * | 2019-10-12 | 2020-03-06 | 平安国际智慧城市科技股份有限公司 | Resource recommendation method, device and storage medium |
Family Cites Families (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8996530B2 (en) * | 2012-04-27 | 2015-03-31 | Yahoo! Inc. | User modeling for personalized generalized content recommendations |
| US20140025490A1 (en) * | 2012-07-17 | 2014-01-23 | Bharathi Shekar | Automated recommendations based on historic location-preference information |
| US20140344102A1 (en) * | 2013-05-18 | 2014-11-20 | Chaya Cooper | Virtual Personal Shopping System |
| US20150120722A1 (en) * | 2013-10-31 | 2015-04-30 | Telefonica Digital Espana, S.L.U. | Method and system for providing multimedia content recommendations |
| CN103745100B (en) * | 2013-12-27 | 2017-01-04 | 浙江大学 | A kind of method of the collaborative filtering recommending of the dominant explicit feedback of project-based mixing |
| US10810616B2 (en) * | 2016-08-17 | 2020-10-20 | Adobe Inc. | Personalization of digital content recommendations |
| CN108470308B (en) * | 2018-03-29 | 2022-04-01 | 百度在线网络技术（北京）有限公司 | Air travel information recommendation method and device |
| CN109064293B (en) * | 2018-08-09 | 2023-04-18 | 平安科技（深圳）有限公司 | Commodity recommendation method and device, computer equipment and storage medium |
| CN109903143A (en) * | 2019-03-27 | 2019-06-18 | 深圳市活力天汇科技股份有限公司 | A kind of flight recommended method based on customer consumption level |
-
2020
- 2020-04-28 CN CN202010349887.0A patent/CN111581505B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110569447A (en) * | 2019-09-12 | 2019-12-13 | 腾讯音乐娱乐科技（深圳）有限公司 | network resource recommendation method and device and storage medium |
| CN110866181A (en) * | 2019-10-12 | 2020-03-06 | 平安国际智慧城市科技股份有限公司 | Resource recommendation method, device and storage medium |
Non-Patent Citations (2)
| Title |
|---|
| Integrating selection-based aspect sentiment and preference knowledge for social recommender systems;Yoke Yie Chen等;Online Information Review;第44卷(第2期);全文 * |
| 基于张量分解和深度学习的混合推荐算法;张家精等;南京大学学报(自然科学);第55卷(第6期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111581505A (en) | 2020-08-25 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105426528B (en) | A kind of retrieval ordering method and system of commodity data | |
| US7805339B2 (en) | Systems and methods for facilitating internet shopping | |
| US10277696B2 (en) | Method and system for processing data used by creative users to create media content | |
| CN111581506B (en) | Flight recommendation method and system based on collaborative filtering | |
| US20160191450A1 (en) | Recommendations Engine in a Layered Social Media Webpage | |
| US20150332373A1 (en) | Method and system for pushing mobile application | |
| US20130046584A1 (en) | Page reporting | |
| US20070027811A1 (en) | Pay-for-access legal research system with access to open web content | |
| US11798051B2 (en) | Configuring food-related information search and retrieval based on a predictive quality indicator | |
| US20090076927A1 (en) | Distinguishing accessories from products for ranking search results | |
| CN103473354A (en) | Insurance recommendation system framework and insurance recommendation method based on e-commerce platform | |
| CN104933100A (en) | Keyword recommendation method and device | |
| CN110163703B (en) | Classification model establishing method, file pushing method and server | |
| US10909571B2 (en) | Visitor identification based on feature selection | |
| US20170155939A1 (en) | Method and System for Processing Data Used By Creative Users to Create Media Content | |
| CN111310032B (en) | Resource recommendation method, device, computer equipment and readable storage medium | |
| US8700648B2 (en) | Context based networking | |
| US10127322B2 (en) | Efficient retrieval of fresh internet content | |
| Chang et al. | Personalized travel recommendation: a hybrid method with collaborative filtering and social network analysis | |
| US20120005021A1 (en) | Selecting advertisements using user search history segmentation | |
| Palomares et al. | Multi-view fuzzy information fusion in collaborative filtering recommender systems: Application to the urban resilience domain | |
| CN111310046A (en) | Object recommendation method and device | |
| CN111581505B (en) | Flight recommendation method and system based on combined recommendation | |
| Niu et al. | Predicting image popularity in an incomplete social media community by a weighted bi-partite graph | |
| Sang et al. | A ranking based recommender system for cold start & data sparsity problem |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |