Skip to main content

Feedback and train

Feedback​

Each time you ask a question the system generate interpretations and computes a confidence score. A list of different interpretations can be shown by clicking on the button at the left of the current interpretation.

note

The confidence is shown in the bar under the interpretation. The system believes the answer is correct if the confidence is bigger than 50% and wrong if it is lower than 50%.

By using the feedback functionality, i.e. by clicking on either the thumbs-up 👍 or thumb-down 👎, the system can learn to choose the right interpretation and correctly compute the confidence.

When asking "What is a cocktail" as a response a list of all cocktails is given, i.e. the system understood that we search for anything of type cocktail which would correspond to "Give me all cocktails".

You can click on the "SPARQL interpretation" icon:

(at the left of the current interpretation) all generated interpretations will be shown:

In particular the second returns general information about "Cocktail". This is the interpretation we are searching for. By clicking on this interpretation we get:

Which gives general information about "Cocktail". By using the feedback functionality (i.e. 👍 or 👎), the system will save the given example.

After asking a set of questions, and by giving feedback you will have created a training set. A training set for the cocktail dataset can be downloaded here:

Training​

You can access the Training page using the button in the status bar (on top of your screen) or in the QA Systems page by clicking on the QA system you are interested in and then the

Training

button.

It can be uploaded from the training page using the "Upload Train" button (available either if you already have feedbacks or not).

On this page you have an overview of how the system performs on the questions you gave feedbacks and some statistics. You can also see a tab name Logs which will show you the history of questions asked.

After clicking on Train the system will create a model that adapts to your dataset:

note

The model will generalize also over questions you didn't ask.

Now you learned how to give feedbacks to the system and how these feedbacks can be used to adapt QAnswer to your QA Systems.

Logs​

This tab is used to see a list of all questions that have been asked to this QA System.

Sharing​

Let's share the QA System with other users, here userB with read rights and userC with write rights. More information about sharing can be found here.

These users can now access the QA System and give feedbacks to the system. As you can notice on the Training page.

note

You can see who gave the feedbacks in the Feedbacks tab. And also see the history of questions asked in the Logs tab to know who asked what.

tip

You can filter your feedbacks and log depending on a column value.

Let's add a filter related to users, click on the three-dot icon to open the menu:

Let's use the Filter option to filter the feedbacks by user.

You can filter in various ways and adapt the filters to your needs.