[
https://issues.apache.org/jira/browse/OPENNLP-842?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17929320#comment-17929320
]
Anish Hiranandani commented on OPENNLP-842:
-------------------------------------------
Hi, I would like to understand the requirements of this story.
> Introduce a Sentiment Quantification component
> ----------------------------------------------
>
> Key: OPENNLP-842
> URL: https://issues.apache.org/jira/browse/OPENNLP-842
> Project: OpenNLP
> Issue Type: New Feature
> Reporter: Mondher Bouazizi
> Priority: Major
> Labels: gsoc, gsoc2016
>
> In addition to the sentiment analysis Component [1], a sentiment quantifier
> is required. In many cases, particularly for long texts, multiple sentiment
> are present. The classification task might be able to detect the most
> dominant sentiment in the text. However, it is as much important to detect
> the other sentiments and attribute different sentiment scores to these
> sentiments.
> Therefore, the objective of this component is to attribute sentiment scores
> after ternary classification: if a text is classified as positive for
> example, the positive sentiment sub-classes (e.g., "love", "happiness",
> "fun", etc.) are attributed different scores showing how much each one of
> them appears in the text. The work [2] presents a good start point, and
> further iteration on the idea are to be made.
> -------------------------------------------------
> [1] OPENNLP-840
> [2] http://www.ieice.org/ken/paper/20160129DbfF/eng/
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
