Natural language generation (NLG)

06/04/2020

What is this? Natural Language Generation (NLG) is a technology that transforms structured data into natural language. NLG is part of the NLP (Natural Language Processing) domain which encompasses software that interprets or produces human language, in either spoken or written form.

WHAT IS THIS?

Natural Language Generation (NLG) is a technology that transforms structured data into natural language. NLG is part of the NLP (Natural Language Processing) domain which encompasses software that interprets or produces human language, in either spoken or written form.

IN A NUTSHELL

NLG: the ability to create context-based, naturally formulated language based on structured data.It turns structured data into written narrative, making data universally understandable.

NLU: Capacity to understand text and meaning of sentences based on natural languages rules in order to convert data into a structured set that a machine can understand and act upon.

In a nutshell

WHAT ARE THE BENEFITS OF NLG?

Natural Language Generation offers significant opportunities to improve operations and the user experience, thereby increasing efficiency:

automation of content generation
delivery of text in a defined format
ability to provide understandable text related to data numeric reporting / database

Reports and data-driven narratives are generated with a reach, depth, and speed not currently possible manually.

As a result, complex reports are faster to produce, with higher accuracy and consistency and it allows systematic analysis of internal reporting or towards clients.

LANGUAGE TECHNOLOGY

Language technology

OUR VISION

NLG is being used to help transform businesses. BNP Paribas Securities Services believes that the technology provides an innovative means to better engage with our clients. Especially where large sources of data are concerned. The implication by creating data driven reporting systems is greater comprehension, not just for the client but also for the employee. We anticipate developing more applications where machines produce easy-to-consume natural narratives at a volume and scale that can better responds to our clients expectations.

WHAT ARE WE DOING

At BNP Paribas Securities Services, we currently produce two client reports using NLG. One such example is the Executive Summary of the MIS Report (MIS stands for Management Information Services). Essentially, the creation of a customisable summary of the MIS client booklet (100+ pages), being the comprehensive data of their global custody activity with information about assets under custody, settlement, corporate actions and income.

The original document is produced as an Excel booklet providing clients with a year’s history of activity. Thanks to NLG, the Executive Summary offers a succinct synopsis of the client’s activity at a glance, covering key alerts and data trends with the option to dive into the detail at any point to gain further insight.

The summary includes text commentary and analysis tailored to each client’s activities. Using well-defined rules, NLG engines transform structured data (key statistics, AuC, settlement data, STP rate, CA volumes, etc.) into meaningful narratives for the client.

INDUSTRY IMPLICATIONS

In general NLG is instrumental in automating the writing of data-driven narratives with varied potential use cases including financial reports, clinical studies, risk and compliance reports
Finance Industry There is an immeasurable amount of data available that could be further leveraged. From set parameters, NLG analysis can identify corresponding data to produce written content and narrative in a given format using understandable terms. Financial reports, regulatory filings, executive summaries and suspicious activity reports, amongst others, are some of the potential use cases NLG can address.

REGULATION

The European Union General Data Protection Regulation (GDPR) requires organisations to be able to explain decisions made by their algorithms. The European Commission published a report on Ethics Guidelines for a Trustworthy Artificial Intelligence in April 2019, announcing it will soon propose legislation for a coordinated European approach on AI. On July 2019, the EU Commission published a factsheet on Artificial Intelligence for Europe which underlines the importance of AI to boost the EU’s competitiveness and ensure trust based on European values, as well as its role in improving people’s lives. This factsheet describes the EU’s role in AI with the financial investments the Commission is planning to make, and gives examples of AI projects conducted by the Commission. Financiers using algorithms and big data might have a greater interest in how the Commission plans to ensure AI remains ethical in services.

At international level, 42 countries (the OECD’s 36 member countries, along with Argentina, Brazil, Colombia, Costa Rica, Peru and Romania) formally adopted the first set of intergovernmental policy guidelines on AI (OECD Recommendation on AI) in May 2019. The Recommendation aims to foster innovation and trust in AI by promoting the responsible stewardship of trustworthy AI while ensuring respect for human rights and democratic values. Subsequently, in June 2019, the G20 Digital Economy Ministers also outlined its commitment to a human-centred approach to AI, publishing a series of G20 AI Principles drawn from the OECD Recommendation on AI.

Other notable government initiatives include setting up AI ethics councils or task forces, and collaborating with other national governments, corporates, and other organizations. Though most of these efforts are still in initial phases and do not impose binding requirements on companies (with GDPR a prominent exception), they signal growing urgency about AI ethical issues.