MarketPsych NLP Engine

An NLP-as-a-service platform for financial text analytics
brochure-image
MarketPsych NLP Engine
An NLP-as-a-service platform for financial text analytics
NLP Engine, also known as SENT (Scoring Engine for Natural Text) is MarketPsych's internal natural language processing (NLP) engine, powering the MarketPsych Analytics products.

NLP Engine is now available via API and on-premises, with client-specific customizations. For further details and a demonstration please contact us.

At A Glance

NLP-as-a-service
MarketPsych’s Scoring Engine for Natural Text (SENT), is an NLP-as-a-service platform for financial analytics. SENT was developed to rapidly process large volumes of unstructured textual data from financial sources such as news, social media, fillings, transcripts, and brokerage research. Clients use SENT output (analytics) to generate standardized insights from their unstructured text. SENT identifies and tags millions of entities, tens of thousands of topics and events, and categorizes financial, ESG, and commodities sentiment. SENT can be deployed on premises or in the cloud. SENT is available on a per-client customization basis. For further details and demonstration please contact us.
Aggregated product sentiment using SENT output

Key Features

Entity Recognition
SENT's AI-driven Named Entity Recognition (NER) system tags millions of entities of over 20 types of entities, such as companies, products, currencies, bonds, ETFs, commodities, and places. It additionally links over 130,000 of those entities to LSEG’s PermID for seamless data integration.
Apple Org (PermID: 4295905573) announces new
iPhone Product in California Region
Topic and Event Insights
SENT covers an extensive profile of 1,000+ topics and 4,000+ events in categories such as finance, environmental issues, politics, law, and technology. Topics are organized in a hierarchical taxonomy for easy navigation.
BYD’s new electric model Topic: EV will be unveiled
Event: EV-Launch in August
Sentiment and Emotion Analysis
Leveraging three fine-tuned roBERTa classifiers, SENT quantifies the sentiment and emotional undertones in text, offering insights into market sentiment, ESG considerations, and commodity market dynamics, along with a spectrum of 10 emotional tones.
Sony’s earnings beat analyst expectations.
Financial sentiment: +98%
Analysts are dissatisfied with the company’s progress toward climate targets.
Disappointment: +26%, ESG sentiment: -98%

Popular client use cases

Real-Time Nowcasting
Feed real-time textual data into our system and receive instantaneous insights, enabling you to make informed decisions quickly.
Research Capabilities
Efficiently navigate through extensive libraries of documents to identify key relationships between entities, market trends, and thematic developments.
Regulatory Monitoring
Analyze regulatory changes with automated monitoring of regulatory texts.
ESG Insights
With coverage of over 200 ESG topics and a sentiment analysis model fine-tuned for ESG discourse, gain nuanced insights from ESG reports
Risk Identification
Automate detecting and monitoring risks within textual data. Perform KYC and monitor negative news with over 300 risk-related topics and events flagged for millions of entities.
Fraud Detection and Prevention
Enhance your fraud detection capabilities by analyzing transaction narratives, communication patterns, and reports.
Customer Insight and Engagement
Understand client preferences and concerns regarding your financial products and services.
Market Sentiment Analysis
Input real-time news, analyst reports, and social media feeds to gauge market sentiment.

Output And Accessibility

SENT outputs its analytics in JSON format, ensuring compatibility and ease of integration with existing systems. The output provides detailed, sentence-by-sentence insights.
Figure 1:
Figure: Depiction of SENT's JSON output
Depiction of SENT’s JSON output
Figure: Installation Workflow
SENT is available for deployment in multiple environments: on-premises, cloud-based, and via API. MarketPsych's tech team provides hosting advice, retrieval and visualization guidance for the detailed analytics.