Claim Verification Assistant with Source Citations
A system that takes a factual claim, retrieves supporting or contradicting evidence from a defined corpus, and reports a verdict with the passages it relied on. The honest version of this problem is bounded: verify against a specific trusted corpus rather than claiming to detect truth in general.
Advanced
Needs a team, real planning, and a supervisor who knows the area.1 semester, 2 to 3 students
5 to ship
3 optional extrasSuggested stack
What you should ship
- Indexed evidence corpus from a defined source set, with provenance recorded per document
- Claim decomposition splitting a compound statement into individually checkable parts
- Evidence retrieval and stance classification labelling each passage as supporting, contradicting or unrelated
- Verdict presentation showing supported, contradicted or insufficient evidence, with citations
- Evaluation on a public fact verification dataset with results compared to a published baseline
If you have time left
- Temporal awareness so a claim true in one year and false later is handled correctly
- Browser extension that checks a selected sentence on any page
- Confidence calibration reported as a reliability diagram
The problem
Automated fact checking is frequently proposed and rarely scoped honestly. Detecting whether an arbitrary statement is true is not achievable. Checking a specific claim against a defined, trusted corpus and reporting what that corpus says is achievable, useful, and a good project.
What you build
Claim decomposition, evidence retrieval, stance classification per passage, and a verdict that is always accompanied by the evidence it rests on. The interface should make the evidence the primary output and the verdict secondary.
Why insufficient evidence is the important verdict
Most claims a system like this receives will not be decidable from its corpus. A system that always produces true or false is wrong most of the time in a way that is hard to notice. Making insufficient evidence a first class, frequently returned answer is the design decision that separates a serious project from a demonstration.
How to evaluate it
Use an established fact verification dataset so your numbers are comparable to published work, and report the three way classification rather than a binary one. Then test on claims drawn from outside the corpus and report how often the system correctly declines.
The hard part
Stance classification. Deciding whether a passage supports or contradicts a claim is subtle, particularly with partial support, hedged language and numbers that nearly match.
Scope warning
Choose your corpus narrowly and state it prominently. A tool that checks claims against one organisation's published statistics is defensible. A tool that claims to check anything is not.
Ideas and guidance, not finished projects
These are project ideas and scoping guidance, published free for students to use as a starting point. I do not build, write, or sell final-year projects, and I do not complete coursework for anyone. Take an idea, make it yours, and build it.
Not sure this one fits you?
I help students pick an idea that matches their skills and their deadline, then scope it down until it is finishable. Guidance only, never done for you.
Stuck on something specific?
Writing only gets you so far. If you want an answer to your situation rather than the general case, book a session and we will work through it together. Every session is free; a few slots open each week.
Follow along
New writing, resources and project ideas land here first.