Generating collections of sets based on user provided annotations
Described are systems and methods for establishing and generating collections of sets that contain object identifiers based on user provided annotations for the object identifiers. A set may include one or more object identifiers and each object identifier may include one or more user provided annotations. Annotations from all object identifiers within a set are processed to form a set profile signature representative of the set. The set profile signatures are then compared between different sets to identify similar sets. Similar sets are included in a collection. Utilizing set profile signatures for a set formed based on user provided annotations provides useful relationships between sets that might otherwise not exist.
1. A computing system, comprising:
one or more processors; and
a memory coupled to the one or more processors and storing program instructions that when executed by the one or more processors cause the one or more processors to at least:
obtain a first plurality of sets, each of the first plurality of sets including one or more object identifiers and a link to at least one other set of the first plurality of sets having a probability of traversal that is greater than a first threshold value;
determine a collection profile signature for a collection of sets, the collection profile signature including at least one of a predefined collection profile signature or a set profile signature associated with at least one set from the plurality of sets;
for each set of the first plurality of sets:
determine, based at least in part on a frequency of a non-random unit group in an annotation associated with the set, a respective set profile signature associated with the set;
determine, based at least in part on the collection profile signature and the respective set profile signature, a respective similarity score representing a similarity between the collection profile signature and the respective set profile signature; and
compare the respective similarity score against a second threshold;
determine, based at least in part on the comparison of the respective similarity scores for each of the first plurality of sets and the second threshold, a second plurality of sets from the first plurality of sets; and
include the second plurality of sets in the collection of sets.
2. The computing system of claim 1 , wherein determining the second plurality of sets includes a determination that the respective similarity scores for each of the second plurality of sets exceeds the second threshold.
3. The computing system of claim 1 , wherein the program instructions, that when executed by the one or more processors, further cause the one or more processors to at least:
determine a second set profile signature associated with a second set of the first plurality of sets;
determine, based at least in part on the collection profile signature and the second set profile signature, a second similarity score representing a second similarity between the collection profile signature and the second set profile signature;
determine that the second similarity score does not exceed the threshold; and
exclude the second set from the collection of sets.
4. The computing system of claim 1 , wherein determining the respective set profile signature for each set of the first plurality of sets includes:
obtaining the annotation associated with the set, wherein the annotation is associated with at least one of a plurality of object identifiers of the set;
identifying the non-random unit group in the annotation;
determining a frequency of the non-random unit group within a plurality of second annotations associated with the first plurality of sets; and
determining, based at least in part on the frequency of the non-random unit group, the respective set profile signature.
5. The computing system of claim 1 , wherein the annotation is provided by a user.
6. The computing system of claim 1 , wherein the program instructions, that when executed by the one or more processors, further cause the one or more processors to at least:
determine a user interest of a user associated with a client device;
determine that the collection of sets corresponds to the user interest; and
cause at least a portion of the collection of sets to be presented on the client device to the user.
7. A computer-implemented method, comprising:
determining a collection profile signature for a collection of sets, the collection profile signature including at least one of a predefined collection profile signature or a set profile signature associated with at least one set from a corpus of sets;
determining, based at least in part on a frequency of a non-random unit group in an annotation associated with a first set, a first set profile signature for the first set;
determining, based at least in part on the collection profile signature and the first set profile signature, a first similarity score representing a similarity between the collection profile signature and the first set profile signature;
determining that the first similarity score exceeds a threshold; and
including the first set in the collection of sets.
8. The computer-implemented method of claim 7 , further comprising:
determining a second set profile signature associated with a second set;
determining, based at least in part on the collection profile signature and the second set profile signature, a second similarity score representing a second similarity between the collection profile signature and the second set profile signature;
determining that the second similarity score exceeds the threshold; and
including the second set in the collection of sets.
9. The computer-implemented method of claim 7 , further comprising:
determining a second set profile signature associated with a second set;
determining, based at least in part on the collection profile signature and the second set profile signature, a second similarity score representing a second similarity between the collection profile signature and the second set profile signature;
determining that the second similarity score does not exceed the threshold; and
excluding the second set from the collection of sets.
10. The computer-implemented method of claim 7 , wherein the first set profile signature is further determined based at least in part on a title of the first set.
11. The computer-implemented method of claim 7 , wherein the first set includes a plurality of object identifiers and determining the first set profile signature for the first set includes:
obtaining the annotation associated with the first set, wherein the annotation is associated with at least one of the plurality of object identifiers;
identifying the non-random unit group in the annotation;
determining a frequency of the non-random unit group within a plurality of second annotations; and
determining, based at least in part on the frequency of the non-random unit group, the first set profile signature for the first set that is representative of the first set.
12. The computer-implemented method of claim 11 , wherein the first set is included in a first plurality of sets and the plurality of second annotations is associated with the first plurality of sets.
13. The computer-implemented method of claim 12 , wherein generation of the first plurality of sets includes:
obtaining a second plurality of sets, each of the second plurality of sets including one or more second object identifiers and a link to at least one other set of the second plurality of sets; and
removing a third plurality of sets from the second plurality of sets based on a probability of traversal associated with the links included in the third plurality of sets to generate the first plurality of sets.
14. The computer-implemented method of claim 13 , wherein the probability of traversal is determined by performing a plurality of random walks between linked sets of the second plurality of sets.
15. The computer-implemented method of claim 7 , further comprising:
obtaining a unit group in the annotation;
determining that a first probability of each unit making up the unit group appearing together is greater than a second probability of each unit making up the unit group appearing separately; and
in response to the determination that the first probability is greater than the second probability, determining that the unit group is the non-random unit group.
16. A computing system, comprising:
one or more processors; and
a memory coupled to the one or more processors and storing program instructions that when executed by the one or more processors, cause the one or more processors to at least:
determine a collection profile signature for a collection of sets, the collection profile signature including at least one of a predefined collection profile signature or a set profile signature associated with at least one set from a corpus of sets; and
for each set of a corpus of sets:
determine, based at least in part on a frequency of a non-random unit group in an annotation associated with the set, a respective set profile signature associated with the set;
determine, based at least in part on the collection profile signature and the respective set profile signature, a respective similarity score representing a similarity between the collection profile signature and the respective set profile signature;
determine that the respective similarity score exceeds a threshold; and
include the set in the collection of sets.
17. The computing system of claim 16 , wherein the program instructions, that when executed by the one or more processors, further cause the one or more processors to at least:
determine a second set profile signature associated with a second set;
determine, based at least in part on the collection profile signature and the second set profile signature, a second similarity score representing a second similarity between the collection profile signature and the second set profile signature;
determine that the second similarity score does not exceed the threshold; and
exclude the second set from the collection of sets.
18. The computing system of claim 16 , wherein determining the respective set profile signature for each set of the corpus of sets includes:
obtaining the annotation associated with the set, wherein the annotation is associated with at least one of a plurality of object identifiers of the set;
identifying the non-random unit group in the annotation;
determining a frequency of the non-random unit group within a plurality of second annotations associated with the corpus of sets; and
determining, based at least in part on the frequency of the non-random unit group, the respective set profile signature.
19. The computing system of claim 16 , wherein the program instructions, that when executed by the one or more processors, further cause the one or more processors to at least:
obtain a first plurality of sets, each of the first plurality of sets including one or more object identifiers and a set link to at least one other set of the first plurality of sets;
determine a probability of traversal associated with each set link; and
generate the corpus of sets from the first plurality of sets by removing sets linked via set links associated with the respective probability of traversal below a second threshold.
20. The computing system of claim 16 , wherein the program instructions, that when executed by the one or more processors, further cause the one or more processors to at least:
obtain a unit group in the annotation;
determine that a first probability of each unit making up the unit group appearing together is greater than a second probability of each unit making up the unit group appearing separately; and
in response to the determination that the first probability is greater than the second probability, determine that the unit group is the non-random unit group.