Recently, Twitter has been used as a citizen-engaging platform by Indian Railway Ministry (IRM) for collecting civic issues. However, due to the manual inspection model, a large percentage of complaints are left unaddressed affecting the credibility of the citizen-sourcing mediums. The existing solutions are for English reports and unable to capture the diverse range of code-mixed languages used in the complaints. In this paper, we developed a multilingual cased version of BERT (mBERT) for automated identification of monolingual and multilingual (aka code- mixed) complaints. The proposed mBERT is a transformers- based model pre-trained on multilingual Wikipedia corpus. In addition to the grievances classification, we also employ a BERT multi-label classifier for labelling the tweets with the issues reported in the complaints. The proposed solution approach obtains an accuracy of 82% with the binary classifier and overall accuracy of 95% with the multi-label model. Additionally, a critical analysis is presented on the classified reports and a location-based analysis to visualise the velocity and veracity of complaints across India.

The fields of business process analysis and process mining (PM) analyze business operations to identify, validate, improve, and automate business processes. Most information about business processes is available as unstructured or semi-structured data in IT systems implementing them. Examples of the information include design documents, XML-format configurations, system logs, and database schema descriptions. The information focuses on the IT algorithms for implementing the processes and does not directly map to the business description of the processes. Advances in Natural Language Processing (NLP) techniques related to semantic tagging, topic modelling, and text classification present opportunities to analyze unstructured data. NLP tasks such as text classification rely on an annotated corpus suitable for the domain. The availability of corpus annotated with implementation-specific tags is a well-known limitation. This paper addresses the challenge of mapping annotations from language and domain-level (generic) descriptions of processes into implementation-specific process data. We present a transfer learning-based approach trained on a corpus of annotated domain-level text and semantic tags. We demonstrate that such a learning technique can effectively annotate implementation-level process information. We further compare the use of state of the art Skip-gram, GloVe, ELMO, and BERT-based learning models in implementing our framework.