Aggregated broad topics

1. A method comprising:

by a computing device, deriving a plurality of input topics based on a content item;

by the computing device, generating a matrix of scores for the input topics according to a first set of cross-indexed topics, wherein each of the scores indicates a degree of similarity between a corresponding one of the input topics and a corresponding one of the first cross-indexed topics;

by the computing device, calculating a total score for each of the cross-indexed topics of the first set based on the scores for the respective cross-indexed topic of the first set across all of the input topics;

by the computing device, selecting one or more of the cross-indexed topics of the first set based on the total scores of the cross-indexed topics of the first set;

wherein generating the matrix of scores comprises generating the first set of cross-indexed topics from a database of topics, wherein each of the first cross-indexed topics is based on an entry in the database of topics; and

wherein the corresponding one of the first cross-indexed topics of the first set comprises a broad topic associated with the corresponding one of the input topics by the database of topics.

2. The method of claim 1 , wherein deriving the plurality of input topics comprises using a topic tagger to identify the plurality of input topics based on the content item.

3. The method of claim 1 , wherein each of the input topics is associated with a confidence value that indicates a degree of confidence in the input topic, and generating the matrix comprises excluding each input topic having a confidence value that does not satisfy a threshold condition.

4. The method of claim 1 , wherein the total score for each of the cross-indexed topics of the first set is calculated based on a sum of the scores for the respective cross-indexed topic of the first set across all of the input topics.

5. The method of claim 1 , further comprising excluding from the selected cross-indexed topics of the first set one or more of the cross-indexed topics of the first set having a total score that does not satisfy a predetermined condition.

6. The method of claim 1 , further comprising ranking the cross-indexed topics of the first set according to the total scores.

7. The method of claim 1 , further comprising normalizing the total scores for the cross-indexed topics of the first set to a first upper bound by identifying a second upper bound and scaling the total scores by a value proportional to a ratio of the first upper bound to the second upper bound.

8. The method of claim 1 , further comprising repeating the generating using the first set of cross-indexed topics as the input topics to generate a second set of cross-indexed topics, wherein each of the scores indicates a degree of similarity between a corresponding one of the first set of cross-indexed topics and a corresponding one of the second cross-indexed topics.

9. The method of claim 1 , further comprising:

by the computing device, generating a matrix of scores for the first set of cross-indexed topics according to a second set of cross-indexed topics, wherein each of the scores indicates a degree of similarity between a corresponding cross-indexed topic of the first set of cross-indexed topics and a corresponding cross-indexed topic of the second set of cross-indexed topics;

by the computing device, calculating a total score for each cross-indexed topics of the second set based on the scores for the respective cross-indexed topic of the second set across all of the first cross-indexed topics; and

by the computing device, selecting one or more of the cross-indexed topics of the second set based on the total scores of the cross-indexed topics of the second set.

10. The method of claim 1 , further comprising:

by the computing device, providing one or more of the cross-indexed topics of the first set and associated total scores to a meta-tagger.

11. One or more computer-readable non-transitory storage media embodying software that is operable when executed to:

derive a plurality of input topics based on a content item;

generate a matrix of scores for the input topics according to a first set of cross-indexed topics, wherein each of the scores indicates a degree of similarity between a corresponding one of the input topics and a corresponding one of the cross-indexed topics of the first set;

calculate a total score for each of the cross-indexed topics of the first set based on the scores for the respective cross-indexed topic of the first set across all of the input topics;

select one or more of the cross-indexed topics of the first set based on the total scores of the cross-indexed topics of the first set;

wherein the software is further operable when executed to generate the matrix of scores by generating the first set of cross-indexed topics from a database of topics, wherein each of the cross-indexed topics of the first set is based on an entry in the database of topics; and

wherein the corresponding one of the cross-indexed topics of the first set comprises a broad topic associated with the corresponding one of the input topics by the database of topics.

12. The media of claim 11 , wherein the software is further operable when executed to derive the plurality of input topics by using a topic tagger to identify the plurality of input topics based on the content item.

13. A system comprising: one or more processors; and a memory coupled to the processors comprising instructions executable by the processors, the processors being operable when executing the instructions to:

derive a plurality of input topics based on a content item;

generate a matrix of scores for the input topics according to a first set of cross-indexed topics, wherein each of the scores indicates a degree of similarity between a corresponding one of the input topics and a corresponding one of the cross-indexed topics of the first set;

calculate a total score for each of the cross-indexed topics of the first set based on the scores for the respective cross-indexed topic of the first set across all of the input topics;

select one or more of the cross-indexed topics of the first set based on the total scores of the cross-indexed topics of the first set;

wherein the processors are further operable when executing the instructions to generate the matrix of scores by generating the first set of cross-indexed topics from a database of topics, wherein each of the cross-indexed topics of the first set is based on an entry in the database of topics; and

wherein the corresponding one of the cross-indexed topics of the first set comprises a broad topic associated with the corresponding one of the input topics by the database of topics.

14. The system of claim 13 , wherein the processors are further operable when executing the instructions to derive the plurality of input topics by using a topic tagger to identify the plurality of input topics based on the content item.