
Can the systems and applications you develop communicate with users? Can they truly recognize text or speech, understand the meaning, and then dynamically generate intelligent responses? That capability, which can enhance efficiency through automation and accelerate decision-making, is in growing demand in some vertical markets. And you can provide it to your users with natural language generation (NLG) technology.
Jay DeWalt, COO of Arria NLG, says although you may hear natural language processing (NLP) and natural language understanding (NLU) used interchangeably with NLG, they differ significantly.
“NLP and NLU process textual data. Virtual assistants hear a question and then respond with an answer from a template or data from a website. It’s all about the questions,” DeWalt explains. “NLG is 180-degrees the opposite. NLG generates language from data and is all about the Answers.”
DeWalt says you may use all three technologies in a conversational stack: NLP and NLU to recognize speech or text and understand the meaning of the questions the user is asking, and then NLG to generate the response.
Arria’s technology takes NLG further. It dynamically generates stories or reports, similar to how a human professional could. Arria accomplishes this with a tool that captures how humans think. When the Arria solution hears or reads data, it can sort, compare, or perform other tasks that the brain does instantaneously. “The cool thing is Arria can do it at massive scale,” he says. “And the data can change, and Arria can generate new language from it dynamically.”
DeWalt says once language is generated, it can be pushed to a presentation layer — such as a dashboard, Word document, speech synthesis, or HTML page — to communicate with users.
NLG Tells the Story
Systems levering this technology can, for example, determine and report insights from a spreadsheet or data set, and when new data is added dynamically generate a new response or report, which can accelerate business decision-making.
DeWalt says early adopters of NLG technology are proving value and ROI:
Financial Planning & Analysis:
Organizations use NLG to generate written (or voice) reports in real time, investor reports, or financial analysis for clients. When integrated with business intelligence dashboards, NLG narratives complement data visualizations with diagnostic explanations in human language that add depth and context to analyses.
Pharmaceutical: NLG is providing value during clinical trials, aggregating and summarizing data from participating physicians, detecting anomalies, and generating final reports. The clock starts on a pharmaceutical patent’s validity at the beginning of the clinical trial, so quickly managing reporting can maximize the company’s market window. NLG also enables more efficient and cost-effective reporting.
Retail: NLG can create content for specific buyer personas, making messaging and promotions more relevant — and effective. For online retailers, NLG can personalize product descriptions for each customer in real time by leveraging internal records, online analytics and other data sets from multiple sources. It can also automate routine tasks so retailers can focus on business growth.
DeWalt says Arria also sees NLG used successfully as a complement to robotic process automation (RPA). For FP&A professionals, RPA-NLG integrations streamline corporate practices of gathering, processing, modeling and NLG finalizes the reporting of data.
RPA platforms process vast amounts of information more quickly than ever before—for example, by automating the ingestion and analysis of both structured and unstructured data with public filings and flagging changes in sentiment on which a research analyst can focus. NLG then processes the data to create sophisticated narratives in a format similar to that written by a human subject matter expert in near real-time.
“Many steps in a workflow process require that a human engages or interacts,” he points out. “So, incorporating NLG can engage the knowledge worker with language.” He adds that NLG-RPA solutions can push out narratives, custom-tailored to specific readers, automatically facilitating communication between robots and humans.
Advice: Choose One Use Case to Start
DeWalt says as soon as Arria’s customers and partners see the potential of their NLG technology, they start imagining uses for it everywhere. “But you have to start somewhere,” he says. “Be focused, choose your first use case and implement it.” This course of action allows you to demonstrate value quickly and then to look for other use cases where it can be applied.
When you reach that point, buckle up. DeWalt comments, “Speed of adoption becomes interesting.”
Arria Connect™, the plug-in used to integrate Arria’s NLG technology with UiPath automations, is now part of UiPath’s ecosystem. For UiPath developers who want to leverage NLG in their automation tasks Arria’s activities are designed to accept either tabular data or JSON data, obtained from an upstream process, into Arria’s NLG Studio platform. Arria’s activities allow users to make API requests to a published Studio project, which then returns a narrative to be consumed by the next UiPath activity.