I recently applied to a new position in my firm and I have gotten 2 assessment tasks. One of them is rather easy, but the second one is giving me some headache.
In mattermost, we have 2 help channels for employees where they can ask questions, and an experienced colleague answers the question. The task now is to extract those questions, categorize them and present the data so it can be used to optimize infobase entries.
My question is:
What tool would you recommend to extract the questions from the channel, sort them and present them? (the goal of the task is to create a plan for that).
I think this mostly depends on the format. If everyone in those channels is answering the questions in threads, then you could rely on the fact that every thread in this channel starts with a question and you could then just get a list of threads for this channel with either the Mattermost API or by running specific queries on your database.
You could also use message actions provided by a small plugin-like app you build to be able to click on a question and add it to your internal infobase then (alongside with the answers, if you wish).
There’s no existing tool to do that and it’s hard to answer this question without knowing the details, but maybe those two suggestions help you already.
Sadly I am actually quite lost here, maybe some additional information will help.
I am working for an online marketplace which sells refurbished products, and the help channels are for the first or second level support (each have their own channel).
In the second level support channel, every message (should) start with the ticket number and the link to the case it is linked to. Afterwards a question is asked and the current moderator answers the question in the thread.
In the first level channel there are no ticket numbers, often there is a link to an email conversation with a customer or just a question after a phone call, but nevertheless a moderator will answer the question in the thread.
Now, for my task I do not have to actually extract the data, I just need a plan how to do it and afterwards categorize und present the data.
If I just describe what I would like to do, it would be something like this:
extract all the questions starting from a certain date (I do not need all of them) and get them in some form where I can start to look through them one by one.
Categorize them in certain topics and if the question would have been avoidable
Sort this data and use it to improve the info database articles regarding the topics
I hope this helps to clarify what I am asking for, I am grateful for any further insights and ideas how to extract the questions.
OK, thanks - but without actually seeing the data I’m not sure if there’s an automated way to extract the questions here.
If I would be assigned this task, I would check the existing data and see if there are any patterns that could be used for automatically extracting the questions. For future questions, I would make sure that they get tagged properly, so the extraction tool can be sure to get the right data. This can be achieved in multiple ways, f.ex. by asking the moderators to react to a question with a specific emoji (you can filter on that then later using the bot’s API) or by using a specific word in the question, which would be an ideal solution for using outgoing webhooks in this channel then.
If f.ex. all questions start with QUESTION:, you could create an outgoing webhook that would send this message to a webservice (by using this triggerword) and you can, in this external webservice, store all these questions and work with them in the database there (not sure if a single question mark could also work as a trigger word, so whenever someone types ? the system identifies that as question and sends it to your webservice).
This webservice could then also search for similar questions before storing it as new question.
I just looked into the outgoing webhooks and saw that they only work in public channels, sadly the 2 help channels are private - as I understand it this won´t work.
Is it possible to do the data extraction with slash commands in those channels, with the same methods you already mentioned?
You’re right, didn’t think about that.
Slash Commands should work, yes - also the Apps framework might be interesting for you if the slash commands don’t satisfy your needs.
Another question is the storage of the data, do you know if it is possible to directly embed the information into an excel sheet? (I also have to provide a concept of how the data is presented and excel would be my prefered choice for that)
At the moment it seems like the questions and the data are manually processed and typed into a google sheet, though that seems like a rather tedious way of putting the information together.
The slash command sends information to a web application which then accepts this data and stores it somewhere. Depending on your coding skills, you can store this information in a local database (sqlite, f.ex.) and work with that and on demand you could then export an excel file from there. The web application could also have a web interface where the user can search through all the relayed questions and flag or tag them.
