Datasaur Gets Continued Funding, Strengthens Data Labeling Platform

Data labeling platform developer startup Datasaur has just posted a new investment of $1 million or equivalent to 14,2 billion Rupiah. This round is still the same as the initial funding previously obtained from GDP Venture. There are several angel investors involved, one of them is Calvin French-Owen as the Co-Founder & CTO of the Segment.

This funding will be used to strengthen the platform's capabilities, including minimizing bias in text labeling. As is known, the data labeling process is one of the crucial aspects in the development of artificial intelligence (AI)-based services, especially in modeling natural language processing (NLP).

Datasaur develop tools to help data labelers work more productively and efficiently. This includes increasing data privacy and security – more often than not, data labeling work is carried out privately outsourced.

"Basically at the moment we deal with all forms of NLP, including entity recognition, parts of speech, document labeling, coreference resolution and parsing dependencies. We've built intelligence into the system to help make labeling more efficient and accurate, enabling companies to manage their entire labeling team on an easy management platform," said Datasaur Founder & CEO Ivan Lee. DailySocial.

Currently the Datasaur team is also participating in the Y Combinator acceleration program for batch Winter 2020 in San Francisco. The company's own base is in California and Indonesia.

NLP is the most popular AI implementation in Indonesia

AI is becoming increasingly popular with the emergence of services capable of automating several business processes. One of the most widely used products is chatbot, corporations are busy using the platform to serve automatic replies on every message given by the customer. Some of them BCA (name chatbot: Vera), Telkomsel (Veronika), BNI (Love) and so on.

Behind the technology chatbot, there are various AI tools applied, one of the most significant is NLP. Its function is to make the computer system understand the language and context written by the user. In fact product chatbot The existing ones still have many shortcomings, including the most fundamental one, namely the lack of vocabulary understanding. The impact on services still feels very rigid, not natural like human-to-human conversations.

One of the results of data labeling is used to train machines (known as the concept of machine learning) in order to have a better understanding of the language, by classifying certain words into predefined groups. Some scenarios are carried out, for example, continuously learning new words conveyed by users.

"Apart from all hyping, AI is a technology that is still being developed. Many companies are looking for best practices in their labeling process. The first generation solution to do is to do outsourcing the whole labeling job. Many companies are building 'Mechanical Turk', but for AI," explained Ivan.

He continued, "Now we see companies identify that quality data is one of the most valuable assets for building and improving AI models. Datasaur is here as a next generation solution, we build software to improve best practices in data labeling, to help develop AI workflows. company."

Along with its development, the market share of AI-based products will continue to increase. Research project its value globally will reach US$390 billion by 2025. For data labeling itself, besides Datasaur, in the global arena there are several other services that can help such as Labelbox, Cloudfactory, and even Google Cloud products are also releasing beta versions for AI Data Lebeling Services.

Data labeling implementation scenario

By understanding the input data, there are many things that can be done. From the existing case studies, Datasaur helps companies to do various things, such as understanding contract documents, transcribing customer service conversations, analyzing product reviews, and detecting fake news.

"Our software has been used to detect and flag suspicious fake news articles by the Indonesian government. A case study with one of our clients showed a 70% increase in labeling efficiency after adopting the Datasaur platform, and we still have more room for improvement," said Ivan.

Currently the data labeling platform has been used by various business verticals, ranging from the financial technology industry, health, customer service, social media to chatbot.