A Kind of Magic: Handle Your Documents Smartly (Gini GmbH)

Agenda

19:00 - 19:15 Uhr: Ankommen
19:15 - 20:15 Uhr: Talk
20:15 - ca. 20:40 Uhr: Questions

Please note: This meetup is in cooperation with the non-commercial organization Gesellschaft für Informatik (GI). Due to its non-commercial character there is no food provided but also no marketing. There is good old Munich water for drinks, however. Usually Smart Data- and GI- talks are in German, this talk is in English.

Access to the building is via a secured elevator. Someone will pick you up and get you to the room. Please just wait downstairs and also note the signs at the door and elevator.

Content:

It has become the common knowledge that Big Data are ubiquitous. It is even more or less clear how to make sense of them. But the bigger the Big Data get, the harder it is to find those pieces of them that are actually valuable and smart. At Gini, we have developed our approach to face the challenge. With the mission to fight paper work and make payments magically simple, we receive the documents from our customers as photos taken with their mobile devices, scans or in any other digitalised format. The documents get into Gini pipeline where the magic happens. In the first step the computer vision module boosts the quality of the image, fixes the direction of the letters on it as well as decides what document we are dealing with (a reminder, an invoice, a remittance slip, etc.). The next in the line is the OCR engine which extracts the text from the image and packs it into the document layout file. The layout XML file is then forwarded to the semantics module, which extracts all the information needed to complete the payment or money transfer such as IBAN, the name of the money recipient, how much money should be transferred, by what date, etc. Different document types are treated differently, i.e., different pieces of information will be extracted from an invoice and from a receipt. We also collect the feedback from our users to constantly improve the system.

Bio: Stalpouskaya, Katya

I have always been fond of Languages and Linguistics which lead me to to studying these subjects as my major at Belarusian State University. While still a student, I had a chance to be one of the developers of Chinese Semantic Engine at Invention Machine. It was life changing experience which brought me to where I am now. After finishing my bachelor, I moved to China to get a Master Degree in Applied and Computational Linguistics from Tsinghua University. Thereafter I moved back to Minsk where I was working as a computational linguist on sentiment analysis of customer reviews at Epam Systems. Afterwards, I came to Munich and was working as a doctoral researcher on the EU Project about the role of mass media in conflict areas at LMU. My main duties were to perform automated text and content analysis. Since January 2017, I work as a Product Owner Semantics at Gini GmbH and am still pursuing my PhD.