Long Context Analysis
Use Nova reasoning models to analyze and synthesize information from extensive documents or datasets.
System Prompt Template
You are an expert {domain} assistant. Answer the question being asked using only the document text provided. Follow these instructions:
1. {instructions}
2. {instructions}
User Prompt Template
Question: {question}
Example
Here we ask Nova to explain the logic behind a conclusion made in the Enron Case
We've also included a download.py script to download the document text. Feel free to use for other documents as well.
Amazon Nova 2 Lite System Prompt
You are an expert {domain} assistant. Answer the question being asked using only the document text provided. Follow these instructions:
1. The full case name in italics (e.g., Skilling v. United States).
2. A citation with the volume number, reporter abbreviation, first page of the case, and pin-point page if applicable (e.g., 561 U.S. 358).
3. The court that decided the case and the year of the decision in parentheses (e.g., (U.S. Supreme Court 2010)).
4. A concise summary of the court's holding or legal conclusion relevant to the question.
5. Any important legal principles or precedents established by the case.
6. Optionally, a brief contextual or procedural background if relevant.
Amazon Nova 2 Lite User Prompt
Document text:
1 Lead Plaintiff states that its motion is based on prior
pleadings that it filed in support of its motion for preliminary
approval of the Plan (#5755, 5756), on the briefing in connection
with Lead Plaintiff’s responses to objections to preliminary
approval of the Plan (#5773, 5774, 5775, and 5776), and on the
first Declaration of Professor Roman L. Weil (expert on plan
allocation
and
securities
violation
damages)(#5794),
[Full document content continues...]
Amazon Nova 2 Lite Sample Response
Response
Analysis of the Two-Tiered Fairness Inquiry and Legal Significance
1. Two-Tiered Fairness Inquiry
The court establishes a two-tiered fairness inquiry for class action settlements:
- Tier 1: Overall Settlement Fairness
The court evaluates whether the settlement agreement between plaintiffs and defendants is fair, reasonable, and adequate under Federal Rule of Civil Procedure 23(e)(1) and (2). This ensures the settlement is not collusive and adequately compensates the class.
- Tier 2: Allocation Agreement Fairness
The court applies the same "fair, reasonable, and adequate" standard to the plan of allocation distributing settlement proceeds among class members. As stated in the opinion:
"The same standard of review applies with as much force to the review of the allocation agreement as it does to the review of the overall settlement between plaintiffs and defendants."
This dual inquiry ensures both the substance of the settlement (e.g., adequacy of recovery) and the manner of distribution (e.g., equitable division among class members) meet rigorous legal standards.
2. Legal Significance of In re Chicken Antitrust Litig.
The court’s citation to In re Chicken Antitrust Litig., 669 F.2d 228 (5th Cir. 1982) is legally significant for three reasons:
- Binding Precedent in the Fifth Circuit:
Chicken Antitrust is a controlling precedent requiring courts to apply the fair, reasonable, and adequate standard to both settlement agreements and allocation plans. The court explicitly adopts this standard, ensuring compliance with circuit law.
- Clarification of Judicial Duty:
The case emphasizes that approving a settlement or allocation plan is not discretionary; courts must conduct a rigorous review to prevent collusion and ensure fairness. As quoted:
"The cardinal rule is that the District Court must find that the settlement is fair, adequate and reasonable and is not the product of collusion."
- Policy Favoring Settlements:
Chicken Antitrust reinforces the strong judicial policy favoring settlements in class actions. The court notes that settlements are compromises, and perfect equality among class members is unrealistic. However, the fairness standard ensures compromises do not become unfair to subclasses.
3. Reconciling Unequal Treatment with Fairness
The court reconciles the principle that "a class action settlement need not necessarily treat all class members equally" with the fair, reasonable, and adequate requirement through differentiated treatment justified by claim-specific factors:
a. Legal Basis for Disparity
The court cites Petruzzi’s, Inc. v. Darling-Delaware Co., 880 F. Supp. 292 (M.D. Pa. 1995):
"Disparate treatment of class members may be justified by a demonstration that the favored class members have different claims or greater damages."
This allows allocations to reflect material differences in:
- Type of Securities (e.g., Enron common stock vs. bonds).
- Timing of Purchases/Sales (e.g., losses incurred during the class period).
- Legal Basis for Recovery (e.g., § 10(b) vs. § 11 claims).
b. Application in the Enron Allocation Plan
The court approves the Plan’s use of two categories and multipliers to achieve fairness:
- Category 1 (e.g., Enron common stock, debt): Receives 95% of proceeds because these securities were central to the fraud and had direct price impacts.
- Category 2 (e.g., securities of Enron-affiliated entities): Receives 5% of proceeds because fraud disclosures had no or indirect impact on their prices.
- Multipliers adjust recoveries for § 11 claims (e.g., tripling recovery for certain notes) to align with statutory damages.
This structure ensures class members are treated differently but fairly, based on objective differences in claim strength and economic loss.
c. Rejection of Windfall Concerns
The court overrules objections (e.g., Fiduciary Counselors) that the Plan creates windfalls:
- Hedging/Arbitrageurs: The Plan excludes windfalls by netting gains/losses only for Enron stock and options (common hedging strategies during the class period).
- Capital Structure Arbitrage: The court notes this strategy was not prevalent during the class period (2001), so isolated cases (e.g., Wilbur Ross’s fund) do not invalidate the Plan.
d. Balancing Competing Interests
The court acknowledges that "no allocation will perfectly satisfy all class members" but upholds the Plan because:
- It was negotiated at arm’s length with no collusion.
- It was reviewed and approved by an independent expert (Professor Roman Weil).
- It considered claim rates, timing, and legal theories to avoid disproportionate windfalls.
Conclusion
The court’s two-tiered fairness inquiry—applying the fair, reasonable, and adequate standard to both the settlement and allocation—ensures comprehensive judicial oversight. By grounding disparities in legitimate claim differences (e.g., security type, legal basis, timing), the court reconciles unequal treatment with fairness, adhering to Chicken Antitrust and other precedents while upholding the policy favoring settlements. The Plan’s structure, validated by expert analysis and objection resolution, exemplifies this balance.
Case Citation:
In re Enron Corp. Sec., Derivative & ERISA Litig., 2008 WL 3148350 (S.D. Tex. Sept. 8, 2008).
API Request
import boto3
import json
# Initialize the Bedrock Runtime client
bedrock_runtime = boto3.client('bedrock-runtime', region_name='us-west-2')
# Define the system prompt
system_prompt = """You are an expert {domain} assistant. Answer the question being asked using only the document text provided. Follow these instructions:
1. The full case name in italics (e.g., Skilling v. United States).
2. A citation with the volume number, reporter abbreviation, first page of the case, and pin-point page if applicable (e.g., 561 U.S. 358).
3. The court that decided the case and the year of the decision in parentheses (e.g., (U.S. Supreme Court 2010)).
4. A concise summary of the court's holding or legal conclusion relevant to the question.
5. Any important legal principles or precedents established by the case.
6. Optionally, a brief contextual or procedural background if relevant."""
# Define the user prompt
user_prompt = """Document text:
1 Lead Plaintiff states that its motion is based on prior
pleadings that it filed in support of its motion for preliminary
approval of the Plan (#5755, 5756), on the briefing in connection
with Lead Plaintiff’s responses to objections to preliminary
approval of the Plan (#5773, 5774, 5775, and 5776), and on the
first Declaration of Professor Roman L. Weil (expert on plan
allocation
and
securities
violation
damages)(#5794),
[Full document content continues...]"""
# Prepare the request
request_body = {
"system": [
{
"text": system_prompt
}
],
"messages": [
{
"role": "user",
"content": [
{
"text": user_prompt
}
]
}
],
"toolConfig": {
"tools": [
{
"systemTool": {
"name": "nova_grounding"
}
},
{
"systemTool": {
"name": "nova_code_interpreter"
}
}
]
},
"additionalModelRequestFields": {
"reasoningConfig": {
"type": "enabled",
"maxReasoningEffort": "low"
}
},
"inferenceConfig": {
"temperature": 0.3,
"topP": 0.9,
"maxTokens": 10000
}
}
# Make the API call
response = bedrock_runtime.converse(
modelId="amazon.nova-2-lite-v1:0",
**request_body
)
# Print the response
print(json.dumps(response, indent=2, default=str))
aws bedrock-runtime converse \
--model-id "amazon.nova-2-lite-v1:0" \
--system '[{"text": "Answer the question being asked using only the document text provided. Follow these instructions:\n\n1. The full case name in italics (e.g., Skilling v. United States).\n2. A citation with the volume number, reporter abbreviation, first page of the case, and pin-point page if applicable (e.g., 561 U.S. 358).\n3. The court that decided the case and the year of the decision in parentheses (e.g., (U.S. Supreme Court 2010)).\n4. A concise summary of the court'\''s holding or legal conclusion relevant to the question.\n5. Any important legal principles or precedents established by the case.\n6. Optionally, a brief contextual or procedural background if relevant."}]' \
--messages file://messages.json \
--additional-model-request-fields '{"reasoningConfig": {"type": "enabled", "maxReasoningEffort": "high"}}' \
--region us-west-2
{
"system": "You are an expert {domain} assistant. Answer the question being asked using only the document text provided. Follow these instructions:\n\n 1. The full case name in italics (e.g., Skilling v. United States).\n 2. A citation with the volume number, reporter abbreviation, first page of the case, and pin-point page if applicable (e.g., 561 U.S. 358).\n 3. The court that decided the case and the year of the decision in parentheses (e.g., (U.S. Supreme Court 2010)).\n 4. A concise summary of the court's holding or legal conclusion relevant to the question.\n 5. Any important legal principles or precedents established by the case.\n 6. Optionally, a brief contextual or procedural background if relevant.",
"messages": [
{
"role": "user",
"content": [
{
"text": "Document text:\n \n 1 Lead Plaintiff states that its motion is based on prior\n pleadings that it filed in support of its motion for preliminary\n approval of the Plan (#5755, 5756), on the briefing in connection\n with Lead Plaintiff\u2019s responses to objections to preliminary\n approval of the Plan (#5773, 5774, 5775, and 5776), and on the\n first Declaration of Professor Roman L. Weil (expert on plan\n allocation \n and \n securities \n violation \n damages)(#5794), \n \n [Full document content continues...]"
}
]
},
{
"role": "user",
"content": [
{
"text": "Question: The court states that the \"fair, reasonable and adequate\" standard applies \"with as much force to the review of the allocation agreement as it does to the review of the overall settlement.\" Analyze how this creates a two-tiered fairness inquiry and explain why the court's citation to In re Chicken Antitrust Litig. is legally significant. How does the court reconcile the principle that \"a class action settlement need not necessarily treat all class members equally\" with the requirement that the allocation be \"fair, reasonable and adequate\"?"
}
]
}
],
"additionalModelRequestFields": {
"reasoningConfig": {
"type": "enabled",
"maxReasoningEffort": "high"
}
}
}