Global segmenting and sentiment analysis based on granular opinion detection

A system, comprising:

• a service hosted by a provider network, wherein the service comprises one or more processors and one or more memories to store computer-executable instructions that, when executed, cause the one or more processors to:
• receive a plurality of documents;
• for a document of the plurality of documents:
• identify a plurality of occurrences of a target entity within the document;
• for different occurrences of the plurality of occurrences of the target entity, determine a sentiment polarity for the target entity;
• based on the sentiment polarity for the target entity for the different occurrences, determine a document-level sentiment polarity for the target entity for the document;
• determine an additional sentiment polarity for the target entity as a multi-document level sentiment polarity for the target entity based on the document-level sentiment polarity for the target entity for the document and at least another document-level sentiment polarity for the target entity for at least another document of the plurality of documents;
• and
• perform one or more of:
• store data indicating at least the multi-document level sentiment polarity for the target entity, wherein the data is available for retrieval by a client, or
• provide, to the client, at least the multi-document level sentiment polarity for the target entity.

A method, comprising:

• performing, by a service implemented by one or more computing devices of a provider network:
• receiving a plurality of documents;
• for a document of the plurality of documents:
• identifying a plurality of occurrences of a target entity within the document;
• for different occurrences of the plurality of occurrences of the target entity, determining a sentiment polarity for the target entity;
• based on the sentiment polarity for the target entity for the different occurrences, determining a document-level sentiment polarity for the target entity for the document;
• determining an additional sentiment polarity for the target entity as a multi-document level sentiment polarity for the target entity based on the document-level sentiment polarity for the target entity for the document and at least another document-level sentiment polarity for the target entity for at least another document of the plurality of documents;
• and
• performing one or more of:
• storing data indicating at least the multi-document level sentiment polarity for the target entity, wherein the data is available for retrieval by a client, or
• providing, to the client, at least the multi-document level sentiment polarity for the target entity.

One or more non-transitory computer-accessible storage media storing program instructions that when executed on or across one or more processors of a provider network cause the one or more processors to implement a service to:

• receive a plurality of documents;
• for a document of the plurality of documents:
• identify a plurality of occurrences of a target entity within the document;
• for different occurrences of the plurality of occurrences of the target entity, determine a sentiment polarity for the target entity;
• based on the sentiment polarity for the target entity for the different occurrences, determine a document-level sentiment polarity for the target entity for the document;
• determine an additional sentiment polarity for the target entity as a multi-document level sentiment polarity for the target entity based on the document-level sentiment polarity for the target entity for the document and at least another document-level sentiment polarity for the target entity for at least another document of the plurality of documents;
• and
• perform one or more of:
• store data indicating at least the multi-document level sentiment polarity for the target entity, wherein the data is available for retrieval by a client, or
• provide, to the client, at least the multi-document level sentiment polarity for the target entity.