Generation and use of literary work signatures reflective of entity relationships

A computer system comprising one or more computing devices, the computer system programmed to implement at least:

• a graph generator that generates a graph representing relationships between entities described in a literary work, the entities including persons represented in the literary work, the graph including data values representing types of said relationships, wherein the graph generator generates the graph by a process that comprises extracting information about the entities and relationships from text of the literary work by applying natural language processing to said text;
• a signature generator that generates a signature of the literary work by a process that comprises identifying a plurality of sub-graphs of the graph, and for individual types of the sub-graphs, determining how many times the respective type of sub-graph occurs in the graph;
• and
• a literary work recommender that determines whether to recommend the literary work to a user who has favorably rated a second literary work at least partly by comparing respective signatures of the literary work and second literary work, wherein the literary work recommender performs said comparison and makes said determination in real time in response to a user action performed during a browsing session of the user, said user action performed subsequent to the generation of said signatures, wherein the literary work recommender, in responding to the user action, compares the signature of the second literary work in real time to pre-generated signatures of a plurality of respective literary works.

Non-transitory computer storage that stores executable program instructions comprising:

• a graph generator module that, when executed, generates a graph representing relationships between entities described in a literary work, the entities including persons represented in the literary work, the graph including data values representing types of said relationships, wherein the graph generator module generates the graph data structure by a process that comprises extracting information about the entities and relationships from text of the literary work;
• a graph transformation module that, when executed, generates a signature of the literary work by a process that comprises identifying a plurality of sub-graphs of the graph, and for individual types of the sub-graphs, determining how many times the respective type of sub-graph occurs in the graph;
• and
• a recommendation module that, when executed, determines, in real time, in response to a user action performed during a browsing session of a user who has favorably rated a second literary work, whether to recommend the literary work to the user at least partly by comparing respective signatures of the literary work and second literary work, wherein the recommendation module is configured to re-use the signatures of the literary work and second literary work to generate literary work recommendations for additional users.

A computer process, comprising:

• generating, in computer storage, a graph data structure that represents relationships between entities described in a literary work, the entities including persons represented in the literary work, the graph data structure including data values representing types of said relationships, wherein generating the graph data structure comprises extracting information about the entities and relationships from text of the literary work by applying natural language processing to said text;
• generating the signature of the literary work, wherein generating the signature comprises identifying a plurality of sub-graphs of the graph data structure, and for individual types of the sub-graphs, determining how many times the respective type of sub-graph occurs in the graph data structure;
• subsequently, responding in real time to an action performed by a user during a browsing session by at least:
• comparing the signature to a signature of a second literary work rated favorably by the user, and
• determining, based at least partly on an outcome of the comparison, whether to recommend the literary work to the user;
• and
• re-using the signatures of the literary work and second literary work to generate literary work recommendations for additional users.

A computer system comprising one or more computing devices, the computer system programmed to implement at least:

• a graph generator that generates a graph representing relationships between entities described in a literary work, the entities including persons represented in the literary work, the graph including data values representing types of said relationships, wherein the graph generator generates the graph by a process that comprises extracting information about the entities and relationships from text of the literary work by applying natural language processing to said text;
• a signature generator that generates a signature of the literary work by a process that comprises identifying a plurality of sub-graphs of the graph, and for individual types of the sub-graphs, determining how many times the respective type of sub-graph occurs in the graph;
• and
• a literary work recommender that determines whether to recommend the literary work to a user who has favorably rated a second literary work at least partly by comparing respective signatures of the literary work and second literary work, wherein the literary work recommender performs said comparison and makes said determination in real time in response to a user action performed during a browsing session of the user, said user action performed subsequent to the generation of said signatures, wherein the literary work recommender is configured to re-use the signatures of the literary work and second literary work to generate literary work recommendations for additional users.

Non-transitory computer storage that stores executable program instructions comprising:

• a graph generator module that, when executed, generates a graph representing relationships between entities described in a literary work, the entities including persons represented in the literary work, the graph including data values representing types of said relationships, wherein the graph generator module generates the graph data structure by a process that comprises extracting information about the entities and relationships from text of the literary work;
• a graph transformation module that, when executed, generates a signature of the literary work by a process that comprises identifying a plurality of sub-graphs of the graph, and for individual types of the sub-graphs, determining how many times the respective type of sub-graph occurs in the graph;
• and
• a recommendation module that, when executed, determines, in real time, in response to a user action performed during a browsing session of a user who has favorably rated a second literary work, whether to recommend the literary work to the user at least partly by comparing respective signatures of the literary work and second literary work, wherein the recommendation module is configured to compare the signature of the second literary work in real time to pre-generated signatures of a plurality of respective literary works in response to said user action.