How to get texts proposed (level 2) by Powertext.ai?

The more the algorithm writes, the less work that is left for you!

Squadra Machine Learning Company

Last Update 3 jaar geleden

Texts proposed

Often referred to as 'Level 2', Powertext.ai is also capable of proposing texts based on the example data that it is trained with. 


Compare this proces to what a human mind is capable of: if you read about ± 100 articles about any product category (such as laptops for example), you will be able to tell some interesting facts about it. However, if you read about ± 1000 articles about laptops, you are becoming an expert on the subject and you are able to tell much more about it. Our algorithm works in the same manner, the more example it 'reads', the better (and more) texts it can propose.


Therefore, at the traingsdata section, example texts can be uploaded and the algorithm will start 'reading' it. This proces shouldn't take too long, if you are waiting for more than 30 minutes, something is probably going wrong.


After 'reading' the product texts, the algorithm is trained and can propose texts (by clicking on the blue 'propose texts' button) within the template section. Of course, the more texts have been uploaded, the more texts can be proposed.


Example texts can be texts from suppliers, texts from competitors, texts that are written for outdated products, etc. The wide range of product texts that can be used has a down side as well: if the example texts are garbage, the output will also be garbage. Therefore, the button 'activate' can be used. In case you notice that your propose texts are lower of quality than before, it might be good to deactivate the last batch of product texts that is uploaded.


Lastly, the format of the example texts need to be in a way that our algorithm can analyze the texts. Comparable to the productdata that is uploaded, the first row should contain column headers. Moreover, two important things need to be taken into account:


1) The column of the example text should be called: 'description'. 

2) All the productfeatures also need to be included within the trainingsfile. 


Therefore, the only difference between a 'trainingsfile' and a 'template file', is the 'description' column that is included in the trainingsfile.


Still need help? Message Us