Tip:
Highlight text to annotate it
X
Our project is titled Identification of Safest Path using Crime Records.
The problem that we wanted to address was finding the safest path to travel in a city.
With the rise in crime activity, personal safety is a big issue, and we wanted to find a path which passes through the least crime prone areas.
This is a particularly useful tool for commuters to avoid crime prone areas
We thought of representing the city by a graph
where the locations are represented by nodes
connected by edges that represent the roads.
We can assign each node and edge a risk
The safest path between two locations is the same as the least risky path between the corresponding nodes.
Given a graph, the least risk or cost path between two nodes can be found by the famous Dijkstra's algorithm.
So we just have to specify the graph.
The structure of the graph depends upon the roads and locations,
and is thus already known.
We just have to assign each node and edge the risk values.
There is no direct way of assigning these risks
Since detailed and up-to-date information is not available.
We decided to use two main sources for collecting data:
Police records and news reports.
We used police records to get a reliable historical value of the risk associated with each area
This was achieved by making a crawler for the Delhi Police website.
Next, we used newspaper reports to obtain updated information.
Mining information from newspaper reports is not straightforward,
because the articles use natural language which
does not rigidly define the type of article or the locations referred inside it.
We used semantic analysis, an automated technique which analyses the meaning of an article
to first identify crime articles, and then a Bayesian model to determine the location referred inside the article.
Using these automated methods we can create a graph with associated risks,
and we can now find the safest path.
The graph is shown with the safest locations coloured more green, and the unsafe locations coloured more red.
The interface that we have created takes the source and destination as input,
and using all the information we have stored in the graph,
it calculates a path which goes through the safest possible areas of the city.
For example, if someone wishes to go from Dwarka to New Friends Colony,
they simply enter the locations,
and the safest path is calculated, and is shown with directions.
The map clearly shows that the shortest path is quite different from the safest path,
which visibly passes through greener or safer areas.