BeSafe

Introduction

It comes as no surprise that people are vulnerable in the national capital. From rape, abduction to murders, vehicle thefts, Delhi is changing its status from the political capital of India to the crime capital. It is topping the charts for all the wrong reasons. In fact, two out of five crimes committed in the country occur right here in Delhi. If we look at data for metro cities only, then Delhi is miles ‘ahead’ of any other big city. 

Many of these locations may be very close to your residence, or places where you travel to on a regular basis. Is there a way to check how safe the place is before visiting it? Is there a way to know what risks you might face while driving through a random neighborhood? Is there a way to know how safe your own locality is? Wouldn’t it be great to know the answers to all these?

To tackle the problem, we came up with an application that uses the information publicly shared on social media and combines them to generate useful inferences from it. The application displays the protection level or crime rates of any selected region within Delhi. The source of inferences is chosen to be credible verified twitter handles that tweet about the crimes in different parts of Delhi. The user would also be able to select the type and severity of crime they wish to see the stats for. The application also provides a well-formatted detailed analysis of the collected tweets along with details of the crime stats on the map. With the application, we also intend to notify the user about potential risks they might face along their routes.

Methodology

Firstly we collected the relevant data. For data collection, we scrapped tweets from twitter handles related to crimes in Delhi. We shortlisted 9 such twitter handles (@DCPEastDelhi, @DCPSEastDelhi, @DcpNorthDelhi, @DCPWestDelhi, @DCPCentralDelhi, @DCPNEastDelhi, @dcpouter, @DCPNewDelhi and @DCPNWestDelhi). All these twitter handles are credible. We scrapped the data from these handles using Tweepy (Twitter API). There were a limited number of tweets collected because of API constraints. A total number of 10,092 tweets were scrapped. These tweets were then filtered and cleaned. 

We also created two lists.  First list contained locations/ localities in New Delhi and the second contained all types of crime. These crimes were divided into 4 main categories : murder, theft, peddling and criminal intent. All the words related to these categories were researched and added to the list.

We then used key word search to find out tweets that contained both location and crime. There were around 456 tweets that were relevant to our project and contained both location and crime type. A .csv file containing tweet, location and crime was generated and passed to the front end.

The Web Application

The web application is designed considering different ways of representing the information to get a clear picture of the crimes happening in Delhi NCR.

Dashboard

The Dashboard section on the Web App consists of 4 ways to represent information extracted from Tweets.

1. Graph showing crimes record sorted by month

The above graph shows the crimes sorted by Month from the year 2020-2021. As seen from the above graph, the highlighted point represents that there were a total 52 crimes in September 2020.

2. Plot showing number of crimes based on categories

The above plot shows the crime data grouped by the Categories.

For eg - The highlighted part in the graph shows that there were a total of 38 Murders from May 2020 to April 2021.

3. Plot showing the top 10 crime hotspots according to tweets

The above bar plot shows the top 10 locations based on the number of crimes. The Tweet text was analyzed to get location data and crime data as well as to know the number of crimes happening in a particular location and then we can group them together.

For eg - 66 Tweets mentioned Sarita Vihar prone to some crime (maybe Theft, Murder, Pickpocketing, etc.).

4. Most recent tweets

In this section, we have introduced a feature to select the crimes based on some interval of time. Specifically, we have introduced the options to select for the last month, 3 months or 6 months.

    i.     Crimes by Last Month

The above graph shows the crimes that have happened in the Last Month. The crimes in the last month show a less number of crimes.

For eg - On 2nd April, 2021, the number of Tweets mentioning some crime were 3.

    ii.    Crimes by Last 3 Months

The above graph shows the crimes that have happened in the Last 3 Months.

For eg - On 17th Feb, 2021, the number of Tweets mentioning some crime were 21.

    iii.    Crimes by Last 6 Months

The above graph shows the crimes that have happened in the Last 6 Months.

For eg - On 31st Dec, 2020, the number of Tweets mentioning some crime were 21.

Map

The next section is a Map section where the app locates the location on the map to give a better understanding of the location and its neighborhood.

The MAP has the following features -

1. The map lets you select the crime sub-category.

2. The Map highlights the locations of the crimes you selected.

For eg - It shows about the “Theft” crime in the below picture.

3. The Map lets you choose the location and shows a drop down menu of the location you type.

4. You can look for a crime in a particular location.

For eg - The below picture shows the number of crimes in Mahipalpur.

Use Cases

The use cases identified by our team (and other kind people giving feedback!) for using the app span multiple scenarios. Some of them are as follows:

Precaution

• Users travelling always have to be wary of thefts and unsafe areas. This app can help  them identify such areas so they can be prepared. For instance, people travelling alone can choose to travel in groups if they are visiting an area which has a high crime rate.

• Shifting to a new location without any knowledge of what the area is like will be a thing of the past. It will also change the way people research for buying or renting property. The app can be a good indicator of whether the locality is notorious for nefarious activities.

• The app could be a provider of timely updates about any area of interest for anyone.

• The app can be used to identify safe spaces for car parking.

Pre-emption and Data analysis

• The app data could be helpful to multiple authorities, including journalists and policemen. The trends which are visualized could help them take appropriate steps to prevent further crime.

• Long term analysis of the data which is collected can be very useful to figure out temporal and spatial trends of crimes, and of very high value to researchers and policemen alike.

Feedback

We identified people from a wide demographic spectrum to provide us with their feedback. This is what they had to say about our app:

• “I can definitely use this before travelling to a new location, to check for crime related activities as I usually have to travel a lot.”

• “Very useful application to beware of car-jackings in a particular area.”

• “Can be used by reporters and the police themselves to check for real time stats.”  

• “A real time graph about crime can be actually used in different ways. From this graph we can actually figure out even the peaks during which serval crime happens. are some crimes that mostly happen to be winter rather than summer? Or around a specific festival?”

• “These kinds of applications are required for the society to be alert.”

• “If you know about a place beforehand, you can make a decision regarding your family members coming along with you before you travel.”

• “Good direction,  positive contribution towards society.”

• “A very useful app, it’ll be great for people looking to rent a home at a relatively safer place.”

Video

You can also refer to the Presentation.

The Team

Acknowledgement

This project was carried out as part of the course Privacy and Security on Online Social Media, under the guidance of Prof. Ponnurangam Kumaraguru, at Indraprastha Institute of Information Technology, Delhi in January-April 2021.