sufficient as per the Railways Parameters. For example, if the complaint is about some
ongoing train, the complaint should have a PNR number, rest all information about the
customer can be taken from the existing software. If the complaint is about some failed
transaction, then the complaint should have some value for user id, mobile number,
transaction number, etc.
c. Identification of actionable tasks
In this part of the research, we aim to categorize the complaints into the categories as used by
the existing software. Some example categories are Divyangjan Facilities, Bed Roll, Staff
Behavior, Cleanliness, Passenger amenties, Coach-Maintenance, Water Availability,
Unreserved Ticketing, Catering & Vending Service to mention a few.
d. Identification of the Division, Region and Department for Redressal
Once the complaint is categorised, it can be assigned to the relevant Division, Region and
Department with minimum human intervention.
9. Novelty of the Project
We have found some of the related researches which work around similar situation. In [Kruspe
et. al, 2021], authors have identified the importance of social media in crisis situations. They
emphasize on automatic analysis of messages to extract crucial information. They present a
review of detection problems based upon three research techniques viz. filtering using keyword
and locations, crowdsourcing and machine learning. In [Preotiuc-Pietro et.al, 2019 ], authors
have presented a linguistic analysis of complaining as a speech act. They have used neural
models of complaints using distant supervision. The authors have focused on identification of
syntactic patterns and linguistic markers specific of complaints.
Railway Complaint Tweets Identification has been taken up in [Akhtar & Beg, 2021]. The
objective of this research paper is to classify tweets into complaint or suggestion.
Taking motivation from such researches, we have refined a novel objective. We focus on
identification the nature of tweet(complaint, suggestion or appreciation). The novelty also lies in
leveraging the power of text analytics for identification of actionable tasks if the tweet is a
complaint. We also identify the incomplete information, i.e. whether the minimum required
information has been provided in the tweet. If the incomplete information is there, prompt the
customers to provide the complete information. We further identify the actionable tasks from the
text analysis of tweet complaints, which can then be assigned to the relevant division, region and
department.
10. Strategy
The practice currently being followed for responding to tweets addressed to Railways is manual
analysis. It is apparent that adding automation to the task will help ease the human effort,
improve response time, and may even allow detection of a crisis situation.
Text analytics is a combination of powerful techniques for automatic identification of crucial
information from raw text. In this research project, raw text consists of the tweets posted on
twitter handles of railways and its officials. The proposed approach will benefit both the
railways(by improving the response time and decreasing the human effort) and the customers (by
providing instant response).
11. Target Audience
1. Twitter Users
All those customers who use twitter for communication with Indian Railways. The twitter users
can benefit in terms of first response because the software plugging will speed up the tweet
processing.
2. Railways Twitter Cell
Since the software plug-in that we propose to build in this project aims at reducing the human
effort involved in the analysis of the tweets. The persons working in the Twitter cell of Indian
Railways will benefit in terms of reduction in volume of data that they currently handle.