Implementing Reasoners in your Docassemble Interviews

Once you have generated a reasoner and installed it on your docassemble server, using the reasoner in your interview requires only a few steps.

Include the IBP_Data Module

You need to include the ibp_data module in order to get access to the object definitions for [DAIBPData] and [DAIBPCase].

---
modules:
  - .ibp_data
---

Include the Formatting For Output

The package comes with javascript and css files which are used to provide a cascading tree interface for explanations. If you wish to display explanations, you need to include the following in your features block:

---
features:
  javascript: docassemble.openlcbr:data/static/list_collapse.js
  css: docassemble.openlcbr:data/static/list_collapse.css

Generate the Required Objects

Once the module is loaded, in the objects block of your reasoner, you must create the following three objects.

objects:
  - reasoner: DAIBPData
  - database: DAStaticFile.using(filename="data/sources/your_reasoner.yml")
  - test_case: DAIBPCase

[DAIBPData] is the reasoner object itself. [DAIBPCase] is the test case format. You would replace “your_reasoner.yml” with the name you gave to your own reasoner file.

Populate the Test Case with Factors

Then, in your interview, you need to add factors to the test_case.factors DAList. Any method can be used to add factors to this list, but the objects added to the list must have a name that is equal to the factor’s ID in your reasoner. For example, the following code adds a factor with an ID of “F1”.

test_case.factors.append("F1")

Initiate the Reasoner

In a mandatory code block, initiate the reasoner by loading the database file.

---
mandatory: True
code: |
  reasoner.load(database)
---

Get a Prediction

Once you are done adding factors to your test case, and the reasoner is initialized, you can obtain the prediction and reasons as follows:

reasons = reasoner.predict(test_case, issue="relationship of interdependence")

The “reasons” variable created above with have a reasons.prediction value which will be either ‘p’, ‘d’, or ‘a’ for Plaintiff, Defendant, or Abstain.

Display Reasons

Displaying the reasoner’s explanation for the prediction in the interview is as simple as making a call to reasons.display_tree().

  ${reasons.display_tree()}