Update README, lock langchain CLI to specific version

README.md

@@ -1,9 +1,22 @@
 # AI models for identifying trigger events in disinformation analysis
 Final Dissertation Submission Repository
 
-## Project Description
+## Abstract
 -- todo --
 
+[Project Presentation](https://jillweynes.github.io/LLMsForDisinformationPrediction-GraphVizBuilt/presentation)
+
+## Generated Database Link and Usage Experiments
+Generated Dataset Link: [https://huggingface.co/datasets/WillJeynes/LLMsForDisinformationAnalysis-Dataset](https://huggingface.co/datasets/WillJeynes/LLMsForDisinformationAnalysis-Dataset)
+
+Graph-Based Dataset Visualisation: [https://jillweynes.github.io/LLMsForDisinformationPrediction-GraphVizBuilt/](https://jillweynes.github.io/LLMsForDisinformationPrediction-GraphVizBuilt/)
+
+Usage Experiments (incl graph visualisation) Source Code: [https://github.com/WillJeynes/LLMsForDisinformationPrediction](https://github.com/WillJeynes/LLMsForDisinformationPrediction)
+
+
+
+# This repository:
+
 ## Solution Diagram
 -- todo --
 
@@ -13,8 +26,6 @@ Final Dissertation Submission Repository
 ## Agent Refinement
 [See agent](/agent/)
 
-## Generated Database Link and Usage Experiments
--- todo --
 
 ## Repository Structure
 ```

agent/README.md

@@ -14,15 +14,19 @@ Experiments modifying pipeline
 Experiments with different model types:
 | Model                         | % Correct | % Change |
 |-------------------------------|----------:|---------:|
-| gpt-5-mini                    | 33        | 0        |
-| gpt-5.4-mini                  | 32.4      | -0.02    |
-| llama3.1:8b-instruct-q4_K_M   | ?         | ?        |
-| qwen3.5:9b                    | 0         | -100     |
+| gpt-5-mini                    | 45.51     |          |
+| gpt-5.4-mini                  | 32.4      |          |
+| gpt-5.4-nano                  | 23.28     |          |
+| gpt-4.1-mini                  | 27.85     |          |
+| gpt-4o-mini                   | 32.47     |          |
+| llama3.1:8b-instruct-q4_K_M   | ?         |          |
+| qwen3.5:9b                    | 0         |          |
 
 %age valid URLS
 | Model                         | Number    | % Age    |
 |-------------------------------|----------:|---------:|
 | gpt-5-mini                    | 22/405    | 5.43     |
 | gpt-5.4-mini                  | 29/278    | 10.43    |
-| llama3.1:8b-instruct-q4_K_M   | ?         | ?        |
-| qwen3.5:9b                    | 0         | 0        |
+| gpt-5.4-nano                  | 6/210     | 2.85     |
+| gpt-4.1-mini                  | 15/269    | 5.57     |
+| gpt-4o-mini                   | 27/287    | 9.407    |

agent/prompts/relation.txt

@@ -1,9 +0,0 @@
-Could the following real-world event:
-###TECLAIM###
-
-Be a trigger for the following disinformation:
-###TITLE###
-
-Respond with "RELATION", followed by : followed by a confidence score (VERYHIGH, HIGH, MEDIUM, LOW, VERYLOW) followed by : followed by the reason. Use no other words, just return the score and reason in format.
-
-Ignore wether the event happened or not, purely consider the likiness of causation

agent/prompts/verify.txt

@@ -1,8 +0,0 @@
-Do the search results cited below
-###TESEARCH###
-Support the idea that the following happened:
-###TECLAIM###
-
-Respond with "CONFIDENCE", followed by : followed by a confidence score (VERYHIGH, HIGH, MEDIUM, LOW, VERYLOW) followed by : followed by the reason. Use no other words, just return the score and reason in format.
-
-Dates can be off by a few days, that would still be valid

run.sh

@@ -5,7 +5,7 @@ set -e
 run_agent () {
     echo "Starting LangGraph agent..."
     cd agent
-    npx @langchain/langgraph-cli dev
+    npx @langchain/langgraph-cli@1.1.17 dev
 }
 
 run_ensemble_service () {

supporting/Wrapper/run.ts

@@ -19,6 +19,9 @@ const MODE = process.env.MODE ?? "claim";
 
 const MAX_CONCURRENCY = 5;
 
+const OFFSET = parseInt(process.env.OFFSET ?? "0", 10);
+const LIMIT = process.env.LIMIT ? parseInt(process.env.LIMIT, 10) : null;
+
 const client = new Client({ apiUrl: API_URL });
 
 
@@ -164,10 +167,19 @@ async function processRecord(record: any): Promise<ResultRecord> {
 async function main() {
   console.log("Reading input file...");
 
-  const records = await loadInputs();
+  const allRecords = await loadInputs();
 
-  console.log(`Loaded ${records.length} records`);
+  console.log(`Loaded ${allRecords.length} records`);
 
+  const records = allRecords.slice(
+    OFFSET,
+    LIMIT !== null ? OFFSET + LIMIT : undefined
+  );
+
+  console.log(
+    `Processing ${records.length} records (offset=${OFFSET}, limit=${LIMIT ?? "∞"})`
+  );
+  
   fs.writeFileSync(OUTPUT_FILE, "", { flag: "a" });
 
   const limit = pLimit(MAX_CONCURRENCY);

supporting/dbkf/fetch.py

@@ -27,7 +27,7 @@ DEFAULT_PARAMS = [
     ("organization", "http://weverify.eu/resource/Organization/3727f7b2aa90ec0716693e5464b28d18"), # StopFake
 ]
 
-NUM_RANDOM_CLAIMS = 200
+NUM_RANDOM_CLAIMS = 2000
 
 INPUT_FILE = "../../data/input.jsonl"
 OUTPUT_FILE = "../../data/claims.json"