The Ideas
These are the ideas that I came up with I think these reflect the brief and integrate Machine Learning well into the ideas. The brief talks about the way cities are becoming more connected the “Smart Cities” idea. There are two sides of this firstly it means that residences of the city can interact and benefit from the increased availability of information and the uses that it has. The important thing in the smart city concept is that the reason to collect the information has to provide a greater benefit than it costs. Example of good uses for smart cities is increased traffic control in city centres to reduce traffic and wait time at peak times.
On the other hand this information is personal and privacy can be an issue, even though business and state organisations say the data will be used in an anonymous capacity this doesn’t mean that one isn’t identifiable from the anonymous data.
A video from the YouTube channel Computerphile highlights this:
This video explains the Anonymisation Problem. This is important to consider when looking at the Concept of smart Cities. I came up with some ideas that use this concept, I think it is particularity interesting as supposedly anonymous data may not actually be anonymous.
The First idea that I wanted to look at was if using a neural network could Identify a person based on a very limited subset of anonymous data. While I was working on this idea the Cambridge Analytics case became public and it was clear that tools that were being used by them could identify people in anonymous data and the supposedly anonymous data was incredibly specific to what each entry had done or been while using Facebook. I decided not to continue with this idea because Cambridge Analytics have already proven something like this is possible.
My Second idea was something similar. I’ve always wondered in books where place are if the Author had been more descriptive in location I could understand the journey that Characters in books took. I decided to look at books based in the real world and not books which existed in different worlds as that obviously wouldn’t work.
The Author John Le Carre writes spy novels which often used real locations in his stories I though this would be a good place to start. I identified several places where I could use Machine Learning. The first place I identified the use of Machine Learning was in Natural Language Processing.
Natural Language Processing is the way in which a machine can interrupt a sentence and identify words as well as sentiment or meaning that the words are making. I thought this would be a good way to identify descriptions of locations in the books. The next thing I wanted to do was once I had extracted the descriptions from the Book was to group them together based on the similarities between the descriptions.
The end goal of this whole idea was to identify locations and map them within associations with each other. In a similar way in which it was done for Lord Of The Rings and The Hobbit on this website.
The main issue I had with this idea was how it was connected with people currently in a City. I thought about using a map so that people could physically explore the real world locations by walking around the City. I didn’t think this impacted enough what data could be used and how easy it was to find something so specific with a small description.
I decided to look at Twitter I found that Tweets could be geo-tagged with a location, but users could also turn this off. So the idea that I came up with was to identify the location that the tweet was talking about if the user had not geo-tagged it. I will cover the exact way this Idea will work in the next blog.
Comments