Automatic event categorization for event ticket network systems
In one embodiment, a method includes accessing event information associated with an event listing, calculating one or more category probabilities for the event listing based in part on the event information associated with the event listing using one or more categorization models, and associating with the event listing the category corresponding to the category probability for each category probability greater than or equal to a threshold probability.
1. A method comprising, by one or more computing devices:
accessing event information associated with an event listing;
calculating one or more category probabilities for the event listing using one or more categorization models, respectively, each categorization model calculating a category probability based at least in part on the event information associated with the event listing, each category probability corresponding to a category, the category probability being equal to the probability that the event listing falls within the category;
for each category probability greater than or equal to a threshold probability, associating with the event listing the category corresponding to the category probability; and
displaying one or more of the categories associated with the event listing to a user.
2. The method of claim 1 , wherein each categorization model includes one or more variables based on: a preset list of one or more features; or a list of one or more values of features being associated with the event listing.
3. The method of claim 1 , wherein one or more categorization models are refined by one or more learning algorithms, the learning algorithms being logistic regression, binary classification, multiclass classification, or maximum likelihood estimation.
4. The method of claim 1 , wherein:
the event information identifies one or more of: a title of the event; a description of the event; a self-identified category of the event; a derived category of the event; a keyword of the event; an instructions for the event; an attendee of the event; an organizer of the event; a venue of the event; a ticket type of the event; a ticket price of the event; or a geographic location of the event; and
calculating the one or more category probabilities is further based at least in part on one or more of: the title of the event; the description of the event; the self-identified category of the event; the derived category of the event; the keyword of the event; the instructions for the event; the attendee of the event; the organizer of the event; the venue of the event; the ticket type of the event; the ticket price of the event; or the geographic location of the event.
5. The method of claim 1 , wherein the category probability for the event listing is a percentile rank equal to the percentage of event listings that have a category probability the same or lower than the event listing.
6. The method of claim 1 , wherein calculating the category probability based at least in part on the event information associated with the event listing comprises:
accessing a list comprising a plurality of features;
comparing the list to the event information;
for each feature present in the event information, calculating a value for the feature; and
calculating the category probability based at least in part on one or more of the values of the features.
7. The method of claim 6 , wherein calculating the category probability based at least in part on the event information associated with the event listing further comprises:
for each feature not present in the event information, calculating a value of for the feature equal to zero.
8. The method of claim 6 , wherein the features present in the event information are represented as a matrix.
9. The method of claim 6 , calculating the one or more category C probabilities f c (z) wherein:
f c ( z )=1/1+ e −z ;
and
z=β 0 +β 1 x 1 +β 2 x 2 + . . . +β k x k , wherein z is the result of a comparison between a set of features associated with an event listing and the weights associated with those features, β is a factor associated with a particular feature, and x is an indication of a particular feature in the event listing.
10. The method of claim 6 , wherein the list comprising a plurality of features is determined by a method comprising:
accessing D training event informations, each training event information being associated with a training event listing;
calculating a term frequency for each term i in the D training event informations; and
for each term i with a term frequency greater than or equal to a threshold term frequency, categorizing the term as a feature.
11. The method of claim 9 , wherein the term frequency for each term i in the D training event informations equals (tf−idf) i,j , wherein:
( tf−idf ) i,j =tf i,j ×idf i ;
tf
i
,
j
=
n
i
,
j
∑
k
n
k
,
j
,
wherein tf i,j is the term frequency of term i in document j, n i,j is the number of times term i appears in document j, and k is the total number of terms; and
idf i =log |D|/|{d : t i εd}|, wherein idf i is the inverse document frequency of term i, and d is the current document.
12. A system comprising: one or more processors; and a memory coupled to the processors comprising instructions executable by the processors, the processors operable when executing the instructions to:
access event information associated with an event listing;
calculate one or more category probabilities for the event listing using one or more categorization models, respectively, each categorization model calculating a category probability based at least in part on the event information associated with the event listing, each category probability corresponding to a category, the category probability being equal to the probability that the event listing falls within the category;
for each category probability greater than or equal to a threshold probability, associate with the event listing the category corresponding to the category probability; and
display one or more of the categories associated with the event listing to a user.
13. One or more computer-readable non-transitory storage media embodying instructions that are operable when executed to:
access event information associated with an event listing;
calculate one or more category probabilities for the event listing using one or more categorization models, respectively, each categorization model calculating a category probability based at least in part on the event information associated with the event listing, each category probability corresponding to a category, the category probability being equal to the probability that the event listing falls within the category;
for each category probability greater than or equal to a threshold probability, associate with the event listing the category corresponding to the category probability; and
display one or more of the categories associated with the event listing to a user.
14. The system of claim 12 , wherein each categorization model includes one or more variables based on: a preset list of one or more features; or a list of one or more values of features being associated with the event listing.
15. The system of claim 12 , wherein one or more categorization models are refined by one or more learning algorithms, the learning algorithms being logistic regression, binary classification, multiclass classification, or maximum likelihood estimation.
16. The system of claim 12 , wherein:
the event information identifies one or more of: a title of the event; a description of the event; a self-identified category of the event; a derived category of the event; a keyword of the event; an instructions for the event; an attendee of the event; an organizer of the event; a venue of the event; a ticket type of the event; a ticket price of the event; or a geographic location of the event; and
calculating the one or more category probabilities is further based at least in part on one or more of: the title of the event; the description of the event; the self-identified category of the event; the derived category of the event; the keyword of the event; the instructions for the event; the attendee of the event; the organizer of the event; the venue of the event; the ticket type of the event; the ticket price of the event; or the geographic location of the event.
17. The system of claim 12 , wherein the category probability for the event listing is a percentile rank equal to the percentage of event listings that have a category probability the same or lower than the event listing.
18. The system of claim 12 , wherein to calculate the category probability based at least in part on the event information associated with the event listing comprises executing instructions to:
access a list comprising a plurality of features;
compare the list to the event information;
for each feature present in the event information, calculate a value for the feature; and
calculate the category probability based at least in part on one or more of the values of the features.
19. The system of claim 18 , wherein to calculate the category probability based at least in part on the event information associated with the event listing further comprises:
for each feature not present in the event information, calculate value of for the feature equal to zero.
20. The system of claim 18 , wherein the features present in the event information are represented as a matrix.