Finance NLP 1.5.0 is out!

We are happy to announce the first release of 2023 – Finance NLP 1.5.0

Finance NLP is a John Snow Lab’s product, launched 2022 to provide state-of-the-art, autoscalable, domain-specific NLP on top of Spark.

With more than 100 models, featuring Deep Learning and Transformer-based architectures, NLP for financial documents includes:

• Annotators to carry out Name Entity Recognition, Relation Extraction, Assertion Status / Understanding Entities in Context, Data Mapping to external sources, Deidentification, Question Answering, Table Question Answering, Sentiment Analysis, Summarization and much more, both training and inference!
• Zero-shot Name Entity Recognition and Relation extraction;
• More than 100 pretrained Deep Learning / Transformer-based models;
• Fully integration with Databricks, AWS or Azure;
• A bunch of notebooks and demos ready to showcase its features.
• Full integration with NLP Labs (former Annotation Lab) for managing your annotation projects and train your financial models in a zero-code fashion.
• Compatiblity with Visual NLP, to combine OCR/Visual capabilities, as Signature Extraction, Form Recognition or Table detection, to Finance NLP.

More than 20 notebooks in Finance NLP 1.5.0

We have updated and improved our more than 20 Financial NLP notebooks. Our Certification training is happening Jan, 2023, where we will be taking a look at all of them and testing some of the features of Finance NLP.

# Finance NLP notebooks

## Splitting, Tokenization, Embeddings
[1.Page_Splitting.ipynb] 
[2.Sentence_Splitting_Tokenization.ipynb] 
[3.Word_Sentence_Embeddings.ipynb] 

## Classification
[4.0.Document_Paragraph_Classification.ipynb] 
[4.1.Training_Financial_Classifiers.ipynb] 

## Named Entity Recognition
[5.0.NER_and_ZeroShotNER.ipynb] 
[5.1.Training_Financial_NER.ipynb] 
[5.2.Financial_NER_Inference_Training.ipynb] 
[5.3.ZeroShot_Financial_NER.ipynb] 

## Relation Extraction
[6.0.Relation_Extraction.ipynb] 
[6.1.Additional_Relation_Extraction_Examples.ipynb] 
[6.2.ZeroShot_Relation_Extraction.ipynb] 
[6.3.Relation_Extraction_Training.ipynb] 

## Assertion Status: Understanding Entities in Context
[7.0.Understand_Entities_in_Context.ipynb] 
[7.1.Training_Financial_Assertion.ipynb] 

## Question&Answering
[8.0.Answering_Questions_Financial_Texts.ipynb] 
[8.1.Automatic_Question_Generation_Financial_Texts.ipynb] 
[8.2.Table_Question_Answering.ipynb] 
[8.3.Finetuning_Table_Question_Answering.ipynb] 

## Normalization and Entity Linking
[9.0.Normalization_with_Entity_Resolution_Edgar.ipynb] 
[9.1.Entity_Resolution_Edgar_unique_IDs_Tickers.ipynb] 
[9.2.Entity_Resolution_NASDAQ.ipynb] 
[9.3.Entity_Resolution_Training.ipynb] 

## Augmentation with external sources with Chunk Mappers
[10.0.Data_Augmentation_with_ChunkMappers.ipynb] 
[10.1.Chunk_Mappers_Training.ipynb] 

## Deidentification
[11.Deidentification.ipynb] 

## Graphs
[80.Financial_Graphs_Neo4j.ipynb] 

## Combining with Visual NLP
[90.0.Financial_Visual_Classification.ipynb] 
[90.1.Visual_and_Textual_Classification.ipynb] 
[90.2.Financial_Visual_NER.ipynb]

Are you willing to join? We still have place for you. Drop us a message to support@johnsnowlabs.com.

2 new Assertion Status models to Understand Entities in Context

• Negation Detection comes to Finance NLP. Use finassertion_negationto understand if an entity, extracted with NER, is negated in the context or not
• Time Assertion gets an improvement with a new round of training and new data: finassertion_time_md

2 new Question-answering models in Finance NLP 1.5.0

We have added finqa_bert and finqa_bert_large to carry out Question Answering on your documents: just ask a question (or generate one!) with natural language and get results from your texts!

Data Augmentation models updated

Our offline-accessible Entity Resolvers (finel_edgar_company_name, finel_edgar_irs) and Chunk Mappers (finmapper_edgar_companyname, finmapper_edgar_irs) for US Sec’s Edgar database have been updated with the information of the last quarter of 2022. Let’s remember what they are useful for:

• Entity Resolvers allow you to normalize strings to an official version or an unique identified from an exernal data source. This is very useful when you try to query a data source, as Edgar, with the name of a company you have extracted with NER. The problem arises with the high variations of the company names you can get in your documents. With Entity Resolution, no matters what you get as ORG NER (g., Cadence, Cadence INC, Cadence Inc., Cadence Incorporated…) all of them will be normalized to Edgar’s official Cadence Inc. and its unique ID (IRS) inEdgar.
• Chunk Mappers allow you to use a normalized string (as the normalized company name using Edgar Entity Resolver) to retrieve information from data sources we make available for you on-premises/offline. For example, by using Edgar Chunk Mappersa Candence ORG, you can obtain information as:

New NER models in Finance NLP 1.5.0

New 8 numerical NER models with up to 139 entities trained on 10-Q documents in xlbr format, extracting the text and tagging only the amounts depending on their nature: assets, contraliabilities, stock equity, debt, expense, income, liability, revenue

• Asset NER: Retrieve entities as cash, cash equivalents, investments, etc. finner_10q_xlbr_lg_asset
• Contraliability NER: Retrieve entities as stock repurchase, repayments of debt, treasure stock acquired, etc. finner_10q_xlbr_lg_contra_liability
• Stock Equity NER: Retrieve entities as stock prices, stock shares, common stock, etc.finner_10q_xlbr_lg_contra_stock_equity
• Debt NER: Retrieve entities as leasing activities, lines of credit, debt interests and instruments, etcfinner_10q_xlbr_lg_debt
• Expenses NER: Retrieve different entities related to expenses, losses, debts and payments.finner_10q_xlbr_lg_expense
• Incomes NER: Retrieve entities as income tax and expense benefits. finner_10q_xlbr_lg_income
• Liabilities NER: Retrieve entities as debts, debt intruments, deferred financial costs, etc. finner_10q_xlbr_lg_liability
• Revenue NER: Retrieve entities as revenues from contracts with our without taxes.finner_10q_xlbr_lg_revenue

How to run

Finance NLP is very easy to run on both clusters and driver-only environments using johnsnowlabs library:

!pip install johnsnowlabs

nlp.install(force_browser=True)
nlp.start()

Fancy trying?

We’ve got 30-days free licenses for you with technical support from our financial team of technical and SME. Just go to john snow labs pricing and follow the instructions!