SENTIMENT_AGGREGATE

Use this procedure to perform sentiment analysis and return a single aggregate sentiment score per document. The aggregate sentiment score is a value between -100 and 100.

You specify search keywords as part of a text query and then identify a sentiment associated with the topics in the document.

The mandatory inputs for this procedure include the name of a text index associated with the document set and the text key, which is a unique identifier that identifies each document. If no sentiment classifier is specified, then the default sentiment classifier is used. The default classifier is only available when using AUTO_LEXER.

If the specified topic keyword is not found within the document, then a sentiment score of zero is returned. If no topic is specified, then the aggregate sentiment score for the entire document is returned.

Note: Avoid using AUTO_LEXER with user-defined classifiers as this may provide inconsistent sentiment scores.

Syntax

SENTIMENT_AGGREGATE(
    index_name IN VARCHAR2,
    textkey IN VARCHAR2,
    topic IN VARCHAR2 default NULL,
    clsfier_name IN VARCHAR2 default NULL,
    ttype IN VARCHAR2 default 'EXACT',
    radius IN NUMBER default 50,
    max_inst IN NUMBER default 5,
    use_saved_copy IN NUMBER default 0
) return NUMBER;

index_name

Specify the name of the CONTEXT index for the text column. This parameter is mandatory.

textkey

Specify the unique identifier (usually the primary key) for the document. The textkey is mandatory and is a single column primary key value.

clsfier_name

Specify the name of the sentiment classifier used to perform sentiment analysis. The maximum length supported for a classifier name is 24 bytes. If you do not specify a classifier name, then the default classifier is used.

topic

Specify the topic for which a sentiment score must be generated for this document. If the topic is not specified, then the sentiment score for the entire document is generated.

ttype

Specify the type of search to be performed for this document:

• EXACT: Indicates that the specified search keyword must be searched in the document. This is the default setting.

• ABOUT: Indicates that the thesaurus must be used to find words that are related to the search keywords.

radius

Specifies the radius of the surrounding text to be analyzed during sentiment classification. The default value is 50.

The exact amount of text used for analysis varies from case to case because Oracle Text attempts to find the best match text segment with respect to nearby topic keywords, word boundaries, and sentence boundaries.

max_inst

Specify the maximum number of instances/occurrences of the topic that must be analyzed. The default value for this parameter is 5.

use_saved_copy

Specify whether to refer to the $D table to fetch the copy of the document and what action to take when the copy of the document is not available in the $D table. The default value of this parameter is zero.

See Also: Oracle Text Application Developer’s Guide for an example of using the SENTIMENT_AGGREGATE procedure