ship.md

To ship

What’s the URL of your prototype?

My product still doesn’t work on the web. What I’ve got, mostly, is some code in my computer that hasn’t been deployed (still learning Express, Node.js, etc). But a more substantiated example of what my product should do, with screenshots, is here.

What is on your backlog for shipping?

My plan is to “ship” the first leg of my product, which is a “convert Facebook posts to text” automator, in a more streamlined fashion. That will be done before unveiling the finished version, that will feed into Watson API. For that I need to:

• Create a Javascript webapp that will take a Facebook Graph API call and output JSON. For now I can only do it through the web version of the Graph (using http://graph.facebook.com/XXXXX?fields=comments.limit(500)), where XXXXX is the number of the post. The “convert JSON to plain text” part I’ve figured it out using Javascript.
• Work the visuals of the page, and create alternative export methods, like .txt or .csv.

What date are you aiming for the launch of this version of your product?

I plan to spend a good part of my vacations trying to figure this part. The idea is to launch the Alpha version in the first week of September, with the Watson integration in the future depending on CUNY support, as I’ll need to hire a developer or increase the limits of Facebook and IBM API calls (that costs money).

How are you going to pitch this product to your community/related organizations?

Do you want to know what type of conversations are happening in your Facebook posts with a glance? To understand, with the click of a button, if they are too emotional, aggressive, or rational and civil? Do you want to take the comments out of Facebook, in text form, to better collect and analyze them? Better yet: use the cutting-edge natural language processing power from IBM Watson to help you do so? I introduce you WatchTheTone.

Once you’ve shipped your current version, how are you going to iterate your product?

My idea is to understand how the app will fit in the workflow of my community — mostly engagement editors and people that analyze the impact of journalism, particularly in nonprofits. I’ve spoken to Michael Keller, co-founder of NewsLynx, and a good project is to make WatchTheTone a Sous Chef of NewsLynx. So people could have the sentiment analysis functionality in the same place as the other metrics.

How do you plan to get feedback from your community?

In the last few days I’ve already demoed my prototype to two people in my community, and I got some nice feedback. I think until my app is fully functional, showing, personally, to people that could potentially use it is a good way to get feedback. Later on, when I get a functional version online, I’ll do e-mail follow-ups with users.

—

To reflect

My scope was:

• just right
• too large
• too small

I was a manager for my product

• yes
• no

I would like to manage this product

• by myself mostly
• with a designer/developer/team

How would you assess the decisions you made early on the class?

The headache was caused by the fact that I changed communities more than a month on the development process. If I had more time, I could iron out my ideas and find easier ways to solve my problems. So in the end I had to show mostly an idea, with a few things that needed to be built (mostly a script to clean and export Facebook JSON) in place.

Did you end up solving the problem you set out to do so? If not, what problem did you solve?

I think I did an interesting first step. The problem I set out to solve was “get a better, cheaper, sentiment analysis”, but in the process, and getting feedback from potential users in my community, I understood that just exporting comments from Facebook to plain text would be very valuable. I think a fully automated sentiment analysis has its limitations, even if it’s powered by the sophisticated Watson API, but it could help to prove some hypothesis. Like, for instance, if the post is “sparking a conversation” or generating a fight.

What roles have you played in developing your product? And what roles do you think you are capable of in future projects?

I did everything, actually, since my developer-friend wasn’t available. And for “everything” I mean: reading a lot of documentation from Facebook API, dozens of Stack Overflow threads of people trying to tackle similar problems, Github Readme.mds and Javascript tutorials.

My initial method, which consumed a lot of days, was forking things on Github that I thought would be similar, but that didn’t end up accomplishing anything. So I had to build my solution from scratch, so it involved writing some lines of code to parse Facebook JSON (which was giving all sorts of errors).

I’m not sure what role I’ll be able to perform in future projects, but the most exciting thing is that I finally, after many years of Code Academy false starts, actually learned some coding, and I’d like to continue doing that. So building the backend of this project will be a nice motivation to continue my study.