English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Unstructured data, especially text, continues to grow rapidly in variousdomains. In particular, in the financial sphere, there is a wealth ofaccumulated unstructured financial data, such as the textual disclosuredocuments that companies submit on a regular basis to regulatory agencies, suchas the Securities and Exchange Commission (SEC). These documents are typicallyvery long and tend to contain valuable soft information about a company'sperformance. It is therefore of great interest to learn predictive models fromthese long textual documents, especially for forecasting numerical keyperformance indicators (KPIs). Whereas there has been a great progress inpre-trained language models (LMs) that learn from tremendously large corpora oftextual data, they still struggle in terms of effective representations forlong documents. Our work fills this critical need, namely how to develop bettermodels to extract useful information from long textual documents and learneffective features that can leverage the soft financial and risk informationfor text regression (prediction) tasks. In this paper, we propose and implementa deep learning framework that splits long documents into chunks and utilizespre-trained LMs to process and aggregate the chunks into vectorrepresentations, followed by self-attention to extract valuable document-levelfeatures. We evaluate our model on a collection of 10-K public disclosurereports from US banks, and another dataset of reports submitted by UScompanies. Overall, our framework outperforms strong baseline methods fortextual modeling as well as a baseline regression model using only numericaldata. Our work provides better insights into how utilizing pre-traineddomain-specific and fine-tuned long-input LMs in representing long documentscan improve the quality of representation of textual data, and therefore, helpin improving predictive analyses.

FETILDA: An Effective Framework For Fin-tuned Embeddings For Long Financial Text Documents