Jobs & News
Browser Plug-In Development for Automatic Handling of Cookie Requests
Taking Information Extraction from OCR to the Next Level
We are thrilled to have received the formal commission to expand our machine-learning-based solution for information extraction from scanned-in personal documents. After eliminating major competitors from the bidding process we can now spend all our attention on driving extraction accuracies towards the 100% mark.
Multilingual Text Generation (NLG) for a Google-funded Start-Up
The events in recent years have highlighted the need for accurate and truthful election forecasting. LangTec is therefore extremely pleased to be able to make a contribution in this sector. Partnering with a Google-funded start-up from Munich (Google News Initiative) we will use our powerful NLG engine TextWriter to produce up-to-date reports on pre-election polls and surveys across numerous countries and languages. Our NLG core technology TextWriter not only generates human-like texts, it also provides deep analytical insights instantaneously.
Great to have you join us, Dr. Christian Betz!
We are super-pleased indeed to have Dr. Christian Betz join us as a senior member of staff. Chris holds a PhD in artificial intelligence and brings to the table many years of experice in software engineering, software architecture and machine learning (ML). This is gonna be really great!
Need Massive Amounts of Training Data for Machine Learning? Come and Meet DataGenerator!
Gone are the days when machine learning (ML) was limited by too few training data. Many use cases today require machine learning algorithms to learn complex patterns from huge amounts of training data. In most use cases, however, these training data are hard to come by, especially when dealing with highly personal or confidential document types such as IDs, insurance contracts or social security cards.
To remedy this, LangTec has created DataGenerator, a customised AI solution for generating massively varied amounts of training data based on a very small number of representative sample documents. DataGenerator permits to generate literally hundreds of thousands of unique document instances based on which even the most data-hungry learning algorithms will have enough to munch on.
Automated Stock Market Reports for the English Market
Exciting New Development Project for a Global Tech Giant
Starting in July 2018 some of our most senior devs will be supporting a global technology giant from Munich in the implementation of an automated analysis of their meeting conversations using automated speech recognition (ASR). The aim of the project is to create a fully automated workflow for passing audio snippets through a semantically enriching speech recognition process.
Machine Learning on Complex Product Data for a Global e-Commerce Giant
Reinforcing our long-standing collaboration with one of this planet’s leading e-commerce providers, we are happy to support their newly founded machine learning team working on the analysis of complex product data. Our contribution is based on our unique expertise mix of machine learning (ML) and natural language processing (NLP).
LangTec Kicks BIG AI’s Butt, Hands Down
Wow … David vs. Goliath, featuring LangTec in the role as David.
One of our financial-sector clients decided to pitch LangTec’s customised information extraction solution against an internationally advertised, high-end commerical semantics engine developed by a global tech superpower. And guess what? Our machine learning solution achieved extraction accuracies double that of Goliath’s electronic superbrain!
And yeah, just in case you were wondering: We’re damn proud of that 🙂
Machine Learning from OCR Data for Big International Bank
Not sure what’s happening in the market place these days but everyone seems to be rediscovering good old optical character recognition (OCR) as a fun thing to play with. OCR involves the recognition of textual elements in image material such as scanned-in documents or screenshots. We have been commissioned with our third project along those lines now. Specifically, we will be using advanced machine learning (ML) and natural language processing (NLP) to extract business-critical information from scanned-in documents with utmost precision.