3 min • 09 June, 2020
Topic modeling is a type of statistical model for discovering the "topics" that occur in a text, like product reviews, customer feedback, or other user-generated content.
By leveraging topic modeling, businesses are able to uncover obscure semantic structures in a text body, revealing key insight to inform business decisions.
Typically, topic modeling is a complex and time-consuming task, encompassing the disciplines of machine learning and advanced data science. Many businesses have neither the time nor the resources for such difficult and laborious processes.
This is where GYANA comes in!
We created this quick, easy-to-use feature in GYANA so anyone can perform topic modeling.
What are your customer reviews telling you?
Do you have user-generated content you want to analyze, but don't know where to begin?
Sign up to GYANA here and follow these eight easy steps to topic modeling
Create a column and label it Topic Modelling. Drag and drop the function TOPIC to the formula editor.
Drop the column labeled Text into the first scalar box and type the number 6 into the second scalar box.
This will fill the Topic Modeling column with the topic most relevant to each sentence of your text.
You can change the number 6 to any value you like – this number defines how many of the most relevant topics you want to study
Add a new chart and connect this chart to your data set.
Drag the column Topic Modeling into the x-axis of this chart, visualizing the count of how many sentences each topic is relevant to.
This is machine learning, performed in just a few clicks!
You can take this further by dropping the column Text onto the y-axis and selecting the COUNT of this column.
Drop the column Sentiment into the Color box and select AVERAGE.
This will display how each topic is mentioned on average throughout your text, either positively or negatively.
If you wish, you can go one step further. Insert another new chart and connect this to the data set, too.
Then, drop Sentence onto the x-axis, and Text onto the y-axis, and select COUNT. Then drag and drop Sentiment into the Color box.
This will visualize how the topic of the text varies in sentiment over time, bringing further depth and insight to your analysis.
1 insightful topic analysis.