Recommendation ranking algorithms that optimize beyond booking

1. A computer implemented method for incorporating multiple objectives in a ranked list of search results provided for display as part of a graphical user interface on a display screen of a computing device, the method comprising:

receiving a search query from a client device;

accessing a set of stored listings for goods or services, and probabilities associated with serving the listings;

defining a serving vector as a probability distribution over the set of listings;

providing the serving vector as input to a multi-objective function;

decomposing the multi-objective function into one or more objective functions, each objective function configured to address a different goal in an overall diversity optimization;

generating a first ranked list of the listings based at least in part on the serving vector that maximizes the decomposed multi-objective function, wherein the serving vector is sorted by probability in a descending order;

iteratively post-processing the serving vector after the first ranked list is returned based on the sorted serving vector, the iterative post-processing comprising:

picking a first listing from the first ranked list;

initializing a set as the null set;

appending the first listing of the first ranked list to the set where reranked items are stored;

picking a second listing out of the first ranked list subtracted by the set by maximizing serving probability and a worst-case diversity measure between the listing to be selected and listings already selected in the set; and

appending the second item to the set and repeating the above steps of the iterative post-processing; and

generating a second ranked list based on the first ranked list and the iteratively post-processed serving vector;

providing the listings for display on the client device according to the order of the second ranked list.

2. The computer implemented method of claim 1 , wherein each of the objective functions comprises a coefficient that governs a relative importance of each objective function in the overall diversity optimization.

3. The computer implemented method of claim 1 , wherein the one or more objective functions include a first objective function that represents a total expected number of bookings given the serving vector.

4. The computer implemented method of claim 3 , wherein the first objective function comprises a first probability of booking of a serving item and a second probability of serving the listing.

5. The computer implemented method of claim 1 , wherein the one or more objective functions includes a second objective function that boosts rankings for the listings of the set with low occupancy rate while maintaining a controllable ranking regression for listings of the set with high occupancy rate.

6. The computer implemented method of claim 5 , wherein the second objective function is summed with the first objective function.

7. The computer implemented method of claim 6 , wherein the concave function is a function of occupancy rate, the concave function being configured to be large when occupancy rate for a listing is low and small when occupancy rate for a listing is high.

8. The computer implemented method of claim 6 , wherein the concave function is a function of at least one of revenue, profit, and cost.

9. The computer implemented method of claim 5 , wherein the second objective function comprises a concave function.

10. The computer implemented method of claim 9 , wherein the concave function includes a linear interpolant between a first epsilon value and a second epsilon value.

11. The computer implemented method of claim 1 , wherein the worst-case diversity measure measures at least one of content diversity and social diversity by performing a comparison between listings.

12. A non-transitory computer-readable medium comprising memory with instructions encoded thereon for incorporating multiple objectives in a ranked list of search results provided for display as part of a graphical user interface on a display screen of a computing device, the instructions, when executed by one or more processors, causing the one or more processors to perform operations, the instructions comprising instructions to:

receive a search query from a client device;

access a set of stored listings for goods or services, and probabilities associated with serving the listings;

define a serving vector as a probability distribution over the set of listings;

provide the serving vector as input to a multi-objective function;

decompose the multi-objective function into one or more objective functions, each objective function configured to address a different goal in an overall diversity optimization;

generate a first ranked list of the listings based at least in part on the serving vector that maximizes the decomposed multi-objective function, wherein the serving vector is sorted by probability in a descending order;

iteratively post-process the serving vector after the first ranked list is returned based on the sorted serving vector, the iterative post-processing comprising:

picking a first listing from the first ranked list;

initializing a set as the null set;

appending the first listing of the first ranked list to the set where reranked items are stored;

picking a second listing out of the first ranked list subtracted by the set by maximizing serving probability and a worst-case diversity measure between the listing to be selected and listings already selected in the set; and

appending the second item to the set and repeating the above steps of the iterative post-processing; and

generate a second ranked list based on the first ranked list and the iteratively post-processed serving vector; and

provide the listings for display on the client device according to the order of the second ranked list.

13. The non-transitory computer-readable medium of claim 12 , wherein each of the objective functions comprises a coefficient that governs a relative importance of each objective function in the overall diversity optimization.

14. The non-transitory computer-readable medium of claim 12 , wherein the one or more objective functions include a first objective function that represents a total expected number of bookings given the serving vector.

15. The non-transitory computer-readable medium of claim 14 , wherein the first objective function comprises a first probability of booking of a serving item and a second probability of serving the listing.

16. The non-transitory computer-readable medium of claim 12 , wherein the one or more objective functions includes a second objective function that boosts rankings for the listings of the set with low occupancy rate while maintaining a controllable ranking regression for listings of the set with high occupancy rate.

17. The non-transitory computer-readable medium of claim 12 , wherein the worst-case diversity measure measures at least one of content diversity and social diversity by performing a comparison between listings.