The purpose of this report is to demonstrate how you can leverage ExtRA output for your analysis.
The sentiment result for each comment includes:
Aspect: extracted nouns from the comment.Topic: semantic grouping of aspects, e.g. actual nouns that people might have used are “heater” and “radiator”. ExtRA will aggregate them together into a topic (e.g., “heating.n.01”). This way you get to see all comments that are about all sorts of heaters (and you can ignore the “.n.01” at the end - this refers to different meanings a word can have. wordnet dictionary).Descriptor: adjective that was used next to an aspect.AdCluster: similar to how we use wordnet to roll up multiple aspects into one, adjective cluster is a way to group similar adjective into one cluster. For example, “good” and “Great” will be combined to “nice”. This reduces dimensionality and makes reasoning slightly easier.SentimentCompound: sentiment of an adjective. However, this sentiment is not always reliable, so it should be thought of as an indication, rather than an accurate indicator.The report only analyses the comments that are in English.
Positives
Negatives
We got in total 420 (40%) positive sentiment, 95 (9%) negative sentiment and 535 (51%) neutral sentiment.
This word map demonstrates what have been talked about and their respective sentiment.
Notes: to increase visibility only aspects with count_aspect > 2 & sentiment_compound ≠ 0 are included in the map; the higher the aspect is located, the more often it is mentioned.
Tips: try clicking on the legend.
Tips: try selecting an area on the plot to zoom in.