Database search system and method of determining a value of a keyword in a search
View Patent ↗Methods of determining values of keywords in an internet search are described. According to one aspect of the invention, a method comprises steps of receiving keywords entered for a plurality of searches; detecting converted transactions associated with the plurality of searches; analyzing the converted transactions; and determining values associated with the keywords based upon the converted transactions. According to other aspects of the invention, methods for recommending subsets of keywords and for recommending keywords based upon converted transactions and click through rates are disclosed. A database search system is also disclosed.
1. A computer-implemented method of determining values of keywords in an internet search comprising:
receiving keywords entered by at least one user for a plurality of search terms;
obtaining information about converted transactions associated with a plurality of searches, the plurality of searches having been performed based on the keywords;
estimating a general transaction probability;
estimating unigram keyword probabilities;
determining conditional probability values associated with the keywords based on the converted transactions;
estimating keyword transaction probabilities based on the general transaction probability, the unigram keyword probabilities, and the conditional probability values; and
generating expected transaction values based on the keyword transaction probabilities, wherein the expected transaction values are generated according to:
E
(
T
❘
k
,
S
)
=
P
(
T
❘
k
,
S
)
*
∑
∀
t
R
(
t
❘
k
,
S
)
C
(
t
❘
k
,
S
)
,
where C(t|k,S) is a non-zero value.
2. The method of claim 1 further comprising periodically updating the conditional probability values.
3. The method of claim 1 further comprising periodically updating the expected transaction values.
4. The method of claim 1 , wherein the expected transaction values comprise average values per order.
5. The method of claim 1 further comprising:
ranking the keywords based on a conditional transaction completion probability; and
providing a recommendation of a subset of the keywords to optimize marketing objectives.
6. The method of claim 1 further comprising estimating return on investment (ROI) values for the keywords based on the expected transaction values.
7. The method of claim 6 , wherein the ROI values are based on the expected transaction values and an expected number of transactions.
8. The method of claim 6 further comprising:
determining a maximum entropy based on the ROI values; and
making recommendations for search terms based on the maximum entropy.
9. The method of claim 6 further comprising ranking the keywords or a subset or subsets of the keywords based on their ROI values.
10. The method of claim 6 further comprising generating a relative ranking of the keywords or a subset or subsets of the keywords based on their relative ROI value to advertisers.
11. A database search system comprising:
a storage element storing keywords entered by at least one user for a plurality of search terms and information related to converted transactions associated with a plurality of searches, the plurality of searches having been performed on the keywords; and
a processing system coupled to the storage element, the processing system operative to perform steps comprising:
estimating a general transaction probability; estimating unigram keyword probabilities;
determining conditional probability values associated with the keywords based on the converted transactions;
estimating keyword transaction probabilities based on the general transaction probability, the unigram keyword probabilities, and the conditional probability values; and
generating expected transaction values based on the keyword transaction probabilities, wherein the expected transaction values are generated according to:
E
(
T
❘
k
,
S
)
=
P
(
T
❘
k
,
S
)
*
∑
∀
t
R
(
t
❘
k
,
S
)
C
(
t
❘
k
,
S
)
,
where C(t|k,S) is a non-zero value.
12. The database search system of claim 11 , wherein the processing system is further operative to perform a step of periodically updating the conditional probability values.
13. The database search system of claim 11 , wherein the processing system is further operative to perform a step of periodically updating the expected transaction values.
14. The database search system of claim 11 , wherein the expected transaction values comprise average values per order.
15. The database search system of claim 11 , wherein the processing system is further operative to perform steps of:
ranking the keywords based upon a conditional transaction completion probability; and
providing a recommendation of a subset of the keywords to optimize marketing objectives.
16. The database search system of claim 11 , wherein the processing system is further operative to perform a step of estimating return on investment (ROI) values for the keywords based on the expected transaction values.
17. The database search system of claim 16 , wherein the ROI values are based on the expected transaction values and an expected number of transactions.
18. The database search system of claim 16 , wherein the processing system is further operative to perform steps of:
determining a maximum entropy based on the ROI values; and
making recommendations for search terms based on the maximum entropy.
19. The database search system of claim 16 , wherein the processing system is further operative to perform a step of ranking the keywords or a subset or subsets of the keywords based on the ROI values.
20. The database search system of claim 16 , wherein the processing system is further operative to perform a step of generating a relative ranking of the keywords or a subset or subsets of the keywords based on their relative ROI value to advertisers.
21. The database search system of claim 16 further comprising a user interface coupled to the processing system, the user interface enabling advertisers to access a recommendation for selecting at least one keyword.
22. The database search system of claim 21 , wherein the recommendations are based on the ROI value for the at least one keyword.
23. A computer-implemented method of recommending search terms based upon converted transactions comprising:
receiving keywords entered by at least one user for a plurality of search terms;
obtaining information about converted transactions associated with a plurality of searches, the plurality of searches having been performed based on the keywords;
estimating a general transaction probability;
estimating unigram keyword probabilities;
determining conditional probability values associated with the keywords based upon the information about the converted transactions;
estimating keyword transaction probabilities based on the general transaction probability, the unigram keyword probabilities, and the conditional probability values;
generating expected transaction values based on the keyword transaction probabilities, wherein the expected transaction values are generated according to:
E
(
T
❘
k
,
S
)
=
P
(
T
❘
k
,
S
)
*
∑
∀
t
R
(
t
❘
k
,
S
)
C
(
t
❘
k
,
S
)
,
where C(t|k,S) is a non-zero value;
estimating return on investment (ROI) values for the keywords based on the expected transaction values and an expected number of transactions;
determining a maximum entropy based on the ROI values;
making recommendations for search terms based upon the maximum entropy; and
periodically updating the conditional probability value and the expected transaction value.