Forum Discussion
Text Analytics - Extract key phrases
Data- I have data containing columns like ID, Category, Description, and Further info. The Category column contains various categories like A, B, C, D, E, etc., while the remaining three columns are text fields.
Aim-My objective is to determine 20 essential phrases that aid in recognizing the IDs meant for category 'A' but haven't been labeled as such. I'm attempting this task utilizing Power BI's Text Analytics feature.
To understand the case more, here is a mockup -
I have a dataset.
| ID | Category | Description | Further info |
| 1 | Comics | It is a case Comics. | It includes marvels, superman,etc. |
| 2 | Comics | It is a case Comics. | It includes spiderman and batman. |
| 3 | Eatables | It is a case Eatables. | It includes fruits, veggies. |
| 4 | Objects | It is a case Objects. | It includes table, chair, spiderman. |
In this case,
For example, key phrases like "spiderman" or "marvel" are part of the "Comics" category. However, looking at the category ''object'', we notice that the key phrase "spiderman" is also present but it is a part of "Comics" category. This indicates a potential mislabeling.
My objective is to identify key phrases that are directly linked to the "Comics" category.
These key phrases will be later used to check if any id which is not categorised as 'comics' but has a huge potential to be one due to presence of key phrases.
Could you kindly guide me through the optimal steps to achieve this? Thank you very much!
2 Replies
thinker_02 , Not very clear, what you want to achieve. Q&A visual should help or check Text Filter visual
Text Filter Slicer and how to search on Multiple columns: https://youtu.be/RbeZRJ3uAZE
Hi, Thanks for replying.
I have a dataset.
ID Category Description Further info 1 Comics It is a case Comics. It includes marvels, superman,etc. 2 Comics It is a case Comics. It includes spiderman and batman. 3 Eatables It is a case Eatables. It includes fruits, veggies. 4 Objects It is a case Objects. It includes table, chair, spiderman. In this case,
For example, key phrases like "spiderman" or "marvel" are part of the "Comics" category. However, looking at the category ''object'', we notice that the key phrase "spiderman" is also present but it is a part of "Comics" category. This indicates a potential mislabeling.
My objective is to identify key phrases that are directly linked to the "Comics" category.
These key phrases will be later used to check if any id which is not categorised as 'comics' but has a huge potential to be one due to presence of key phrases.
