2. Select Your Data

Accern Collection

The Accern Collection contains various data sources out-of-the-box. Once they are part of your deployment, you can pick from the following sources:

Public news and blogs are composed of publicly available information throughout the web. Our crawlers go through billions of websites and generate hundreds of thousands of signals related to financial services everyday.

• You can click the “Configure” button to choose a different language from over 100 languages. This will allow you to get public news & blogs content from local publications in a native language. The content will be translated into English before the NLP is run on top.

Dow Jones Newswires (Accern News Analytics - Powered by Dow Jones). This feed is composed of a select set of Dow Jones Newswires that we have access to scrape and then quantify using our analytics (Global Equities, FX, Fixed Income, Energy and Commodities). There are low levels of noise since these Newswires have been curated by Dow Jones

EDGAR SEC 10K & 10Q Filings are scraped from the EDGAR database, allowing us to contextualize each section of these filings within a CSV/API and apply our analytics.

Custom Collection

If you have your own alternative datasets or internal data that you would like to import into Accern, you can choose “Custom Data Sources.” From here, you can connect various sets of unstructured data to be analyzed alone or in conjunction with a data source from the Accern Collection.

After clicking on the “Upload Custom Data” tile, a pop up will appear with a dropdown menu with four selections: JSON, URL, Fileshare, Postgres.

JSON/URL

1. If you choose JSON or URL, you will be generated a URL as your submit endpoint and a token as your authorization value once confirmed. Name your data source, confirm and a new tile with the name you chose will appear as a data source.

2. Using Postman, you can enter your custom data source’s API URL (submit endpoint) and change GET to POST. Next, underneath “Headers” you can add in the provided token and choose “authorization”.

3. Input your content into the body as a raw JSON and click send. For the JSON format required to POST the data, please refer to the format under the details of your custom data tile.

4. Once received, the content that was sent will appear in your use case with selected configurations.

You can only push JSON content or Article URLs into your data tile after you have created it and launched your real-time feed. Once your feed is connected, you can follow the next steps to push your data into the system.

Fileshare

Unlike a JSON or URL custom data tile, your data already exists in a specific directory so you can set up your Fileshare custom data tile without having to deploy the model first.

If you choose Fileshare, you will be greeted with a few fields to fill out.

1. File Type - choose your file type between: CSV, JSON, XML, AVRO

2. Directory Path - define where your files currently reside, such as /data/folder-x/folder-y/final_directory

3. Doc Title - map your data field that corresponds to a title.

4. Doc Text - map your data field that corresponds to the text or content.

5. Harvested At - map your data field that corresponds to your modification or harvested date.

6. Published At - map your data field that corresponds to the content published date.

7. Provider ID - map your data field that corresponds to an ID.

Once filled out and confirmed, you can turn the Fileshare custom data tile on, follow the remaining steps to configure your model and deploy. Once deployed, your model will pull the data from your specified Directory Path.