Systems and methods for on-demand services

1. A system configured to provide an online to offline service to a user, comprising:

at least one storage medium including a set of instructions; and

at least one processor in communication with the at least one storage medium, wherein when executing the set of instructions, the at least one processor is directed to:

obtain an online transportation service request including a target address query via an application being executed on a user terminal associated with the user;

determine at least one of a prefix, a key term, or a phrase in the target address query;

determine a plurality of candidate points of interest (POIs) based on the prefix, the key term, or the phrase;

identify one or more target POIs based on the plurality of candidate POIs by using a trained identification model, which is configured to provide a correlation probability for each of the one or more target POIs with the target address query;

determine a target correlation threshold associated with the trained identification model based on a precision-recall relationship established based on a plurality of historical transportation trip records and the trained identification model;

determine whether a number count of eligible target POIs of the one or more target POIs is less than a count threshold, wherein each of the eligible target POIs has a correlation probability with the target address query larger than the target correlation threshold;

obtain one or more supplementary POIs associated with the target address query from a third party in response to the determination that the number count of the eligible target POIs of the one or more target POIs is less than the count threshold;

add the one or more supplementary POIs to the one or more target POIs;

rank the one or more target POIs and the one or more supplementary POIs to produce ranked target POIs and ranked supplementary POIs based on correlation probabilities corresponding to the one or more target POIs and the one or more supplementary POIs;

transmit the ranked target POIs and the ranked supplementary POIs to the user terminal to be displayed via a user interface on the user terminal;

select, by the user, a POI from the ranked target POIs and the ranked supplementary POIs as a service location; and

distribute the service location to a plurality of transportation service providers.

2. The system of claim 1 , wherein the trained identification model is determined with a training process, the training process comprising:

obtaining a plurality of first historical transportation trip records, wherein each of the plurality of first historical transportation trip records includes a first address query from a user, one or more first POIs associated with the first address query, and a first PO selected by the user from the one or more first POIs as a service location of the transportation trip record;

determining a plurality of first samples including a plurality of first positive samples and a plurality of first negative samples, wherein each of the plurality of first positive samples includes the first address query and the selected first POI and each of the plurality of first negative samples includes the first address query and one of the one or more first POIs other than the selected first POI; and

determining the trained identification model based on a preliminary identification model, the plurality of first positive samples, and the plurality of first negative samples.

3. The system of claim 2 , wherein the determining of the trained identification model based on the preliminary identification model, the plurality of first positive samples, and the plurality of first negative samples includes:

extracting feature information of each of the plurality of first samples;

determining a plurality of sample correlation probabilities corresponding to the plurality of first samples based on the preliminary identification model and the feature information;

determining whether the plurality of sample correlation probabilities satisfy a preset condition; and

designating the preliminary identification model as the trained identification model in response to the determination that the plurality of sample correlation probabilities satisfy the preset condition.

4. The system of claim 3 , wherein the feature information of each of the plurality of first samples includes at least one of a first frequency that the first POI was selected as service locations in the plurality of first historical transportation trips, a second frequency that the first POI was transmitted to the users in the plurality of first historical transportation trips, or a similarity between the first address query and the first POI.

5. The system of Tim 1 , wherein the trained identification model includes a binary classification tree model.

6. The system of claim 1 , wherein to obtain the target correlation threshold associated with the trained identification model, the at least one processor is further directed to:

obtain the plurality of historical transportation trip records;

establish the precision-recall relationship based on the plurality of historical transportation trip records and the trained identification model;

obtain a reference precision value or a reference recall value; and

determine the target correlation threshold based on the reference precision value or the reference recall value and the precision-recall relationship.

7. The system of claim 6 , wherein to establish the precision-recall relationship based on the plurality of historical transportation trip records and the trained identification model, the at least one processor is further directed to:

determine a plurality of actual positive examples and a plurality of actual negative examples based on the plurality of historical transportation trip records;

determine a plurality of predicted positive examples and a plurality of predicted negative examples based on the plurality of historical transportation trip records and the trained identification model; and

establish the precision-recall relationship based on the plurality of actual positive examples, the plurality of actual negative examples, the plurality of predicted positive examples, and the plurality of predicted negative examples.

8. The system of claim 6 , wherein

any point on the precision-recall relationship corresponds to a precision value, a recall value, and a candidate correlation threshold, wherein

the candidate correlation threshold is used to classify a plurality of samples as predicted positive examples and predicted negative examples based on correlation probabilities corresponding to the plurality of samples determined by the trained identification model; and

the precision value and the recall value are determined based on the predicted positive examples and the predicted negative examples; and

to determine the target correlation threshold based on the reference precision value or the reference recall value and the precision-recall relationship, the at least one processor is further directed to:

determine a candidate correlation threshold corresponding to the reference precision value or the reference recall value on the precision-recall relationship as the target correlation threshold.

9. The system of claim 6 , wherein

the reference precision value is a precision value of a previously used model by the system; or

the reference recall value is a recall value of the previously used model by the system.

10. A method implemented on a computing device having at least one processor, at least one storage medium, and a communication platform connected to a network, the method comprising:

obtaining an online transportation service request including a target address query via an application being executed on a user terminal associated with the user;

determining at least one of a prefix, a key term, or a phrase in the target address query;

determining a plurality of candidate points of interest (POIs) based on the prefix, the key term, or the phrase;

identifying one or more target POIs based on the plurality of candidate POIs by using a trained identification model, which is configured to provide a correlation probability for each of the one or more target POIs with the target address query;

determining a target correlation threshold associated with the trained identification model based on a precision-recall relationship established based on a plurality of historical transportation trip records and the trained identification model;

determining whether a number count of eligible target POIs of the one or more target POIs is less than a count threshold, wherein each of the eligible target POIs has a correlation probability with the target address query larger than the target correlation threshold;

obtaining one or more supplementary POIs associated with the target address query from a third party in response to the determination that the number count of the eligible target POIs of the one or more target POIs is less than the count threshold;

adding the one or more supplementary POIs to the one or more target POIs;

ranking the one or more target POIs and the one or more supplementary POIs to produce ranked target POIs and ranked supplementary POIs based on correlation probabilities corresponding to the one or more target POIs and the one or more supplementary POIs;

transmitting the ranked target POIs and the ranked supplementary POIs to the user terminal to be displayed via a user interface on the user terminal;

selecting, by the user, a POI from the ranked target POIs and the ranked supplementary POIs as a service location; and

distributing the service location to a plurality of transportation service providers.

11. The method of claim 10 , wherein the trained identification model is determined with a training process, the training process comprising:

obtaining a plurality of first historical transportation trip records, wherein each of the plurality of first historical transportation trip records includes a first address query from a user, one or more first POIs associated with the first address query, and a first POI selected by the user from the one or more first POIs as a service location of the transportation trip record;

determining a plurality of first samples including a plurality of first positive samples and a plurality of first negative samples, wherein each of the plurality of first positive samples includes the first address query and the selected first POI and each of the plurality of first negative samples includes the first address query and one of the one or more first POIs other than the selected first POI; and

determining the trained identification model based on a preliminary identification model, the plurality of first positive samples, and the plurality of first negative samples.

12. The method of claim 11 , wherein the determining of the trained identification model based on the preliminary identification model, the plurality of first positive samples, and the plurality of first negative samples includes:

extracting feature information of each of the plurality of first samples;

determining a plurality of sample correlation probabilities corresponding to the plurality of first samples based on the preliminary identification model and the feature information;

determining whether the plurality of sample correlation probabilities satisfy a preset condition; and

designating the preliminary identification model as the trained identification model in response to the determination that the plurality of sample correlation probabilities satisfy the preset condition.

13. The method of claim 12 , wherein the feature information of each of the plurality of first samples includes at least one of a first frequency that the first POI was selected as service locations in the plurality of first historical transportation trips, a second frequency that the first POI was transmitted to the users in the plurality of first historical transportation trips, or a similarity between the first address query and the first POI.

14. The method of claim 10 , wherein the trained identification model includes a binary classification tree model.

15. The method of claim 10 , wherein the obtaining of the target correlation threshold associated with the trained identification model includes:

obtaining the plurality of historical transportation trip records;

establishing the precision-recall relationship based on the plurality of historical transportation trip records and the trained identification model;

obtaining a reference precision value or a reference recall value; and

determining the target correlation threshold based on the reference precision value or the reference recall value and the precision-recall relationship.

16. The method of claim 15 , wherein the establishing of the precision-recall relationship based on the plurality of historical transportation trip records and the trained identification model includes:

determining a plurality of actual positive examples and a plurality of actual negative examples based on the plurality of historical transportation trip records;

determining a plurality of predicted positive examples and a plurality of predicted negative examples based on the plurality of historical transportation trip records and the trained identification model; and

establishing the precision-recall relationship based on the plurality of actual positive examples, the plurality of actual negative examples, the plurality of predicted positive examples, and the plurality of predicted negative examples.

17. A non-transitory computer readable medium, comprising executable instructions that, when executed by at least one processor, directs the at least one processor to perform a method, the method comprising:

obtaining an online transportation service request including a target address query via an application being executed on a user terminal associated with the user;

determining at least one of a prefix, a key term, or a phrase in the target address query;

determining a plurality of candidate points of interest (POIs) based on the prefix, the key term, or the phrase;

identifying one or more target POIs based on the plurality of candidate POIs by using a trained identification model, which is configured to provide a correlation probability for each of the one or more target POIs with the target address query;

determining a target correlation threshold associated with the trained identification model based on a precision-recall relationship established based on a plurality of historical transportation trip records and the trained identification model;

determining whether a number count of eligible target POIs of the one or more target POIs is less than a count threshold, wherein each of the eligible target POIs has a correlation probability with the target address query larger than the target correlation threshold;

obtaining one or more supplementary POIs associated with the target address query from a third party in response to the determination that the number count of the eligible target POIs of the one or more target POIs is less than the count threshold;

adding the one or more supplementary POIs to the one or more target POIs;

ranking the one or more target POIs and the one or more supplementary POIs to produce ranked target POIs and ranked supplementary POIs based on correlation probabilities corresponding to the one or more target POIs and the one or more supplementary POIs;

transmitting the ranked target POIs and the ranked supplementary POIs to the user terminal to be displayed via a user interface on the user terminal;

selecting, by the user, a POI from the ranked target POIs and the ranked supplementary POIs as a service location; and

distributing the service location to a plurality of transportation service providers.

18. The non-transitory computer readable medium of claim 17 , wherein the trained identification model is determined with a training process, the training process comprising:

obtaining a plurality of first historical transportation trip records, wherein each of the plurality of first historical transportation trip records includes a first address query from a user, one or more first POIs associated with the first address query, and a first POI selected by the user from the one or more first POIs as a service location of the transportation trip record;

determining a plurality of first samples including a plurality of first positive samples and a plurality of first negative samples, wherein each of the plurality of first positive samples includes the first address query and the selected first POI and each of the plurality of first negative samples includes the first address query and one of the one or more first POIs other than the selected first POI; and

determining the trained identification model based on a preliminary identification model, the plurality of first positive samples, and the plurality of first negative samples.

19. The non-transitory computer readable medium of claim 18 , wherein the determining of the trained identification model based on the preliminary identification model, the plurality of first positive samples, and the plurality of first negative samples includes:

extracting feature information of each of the plurality of first samples;

determining a plurality of sample correlation probabilities corresponding to the plurality of first samples based on the preliminary identification model and the feature information;

determining whether the plurality of sample correlation probabilities satisfy a preset condition; and

designating the preliminary identification model as the trained identification model in response to the determination that the plurality of sample correlation probabilities satisfy the preset condition.

20. The non-transitory computer readable medium of claim 19 , wherein the feature information of each of the plurality of first samples includes at least one of a first frequency that the first POI was selected as service locations in the plurality of first historical transportation trips, a second frequency that the first POI was transmitted to the users in the plurality of first historical transportation trips, or a similarity between the first address query and the first POI.