OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
AI Evaluation & Observability

Factsumm

Factsumm is an open-source Python toolkit for scoring the factual consistency of abstractive summaries against their source articles. It can extract and compare fact triples from both texts and presents additional evaluation outputs in its example, including question-answer-based scores, ROUGE and BERTScore.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Factsumm — OpenViST project profile in AI Evaluation & Observability.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use Factsumm when you have both a source article and its generated abstractive summary and want scores or extracted facts to help inspect whether the summary reflects the source.

Who it is for

  • Developers building abstractive summarization workflows
  • Researchers or practitioners comparing generated summaries with source texts

Documented uses

  • Compare a generated summary with its source article
  • Extract facts from source and summary texts to examine differences
  • Add a factual-consistency check to a summarization evaluation workflow
Before you choose

Requirements & limits

The public repository identifies the project as open source under the Apache-2.0 license. The README specifies Python 3.8 or later and installation through pip or from source.

  • The documented interface takes source and summary text in Python code; it is not presented as a hosted service.
  • The example exposes fact differences as well as a factual-consistency score, with additional QA, ROUGE and BERTScore outputs.
  • The README documents CPU as the default device and shows cuda for GPU use.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Install FactSumm and score a summary

Install FactSumm in Python and compare a source article with its generated summary for factual consistency.

Before you start

  • Python 3.8 or later

Steps

  1. Install the package with the command below.

    Documented command
    pip install factsumm

  2. Import FactSumm and create an instance: from factsumm import FactSumm followed by factsumm = FactSumm().

  3. Provide the source article and generated summary as strings, then call factsumm(article, summary, verbose=True) to inspect the consistency results.

Expected outcome

FactSumm returns a factual-consistency score and, with verbose output, displays extracted entities and facts and their differences.

Guide limitations

  • The README documents this example for a single article-summary pair; it does not provide a batch-evaluation procedure.
  • The documented GPU option is to pass device="cuda"; CPU is the default.
Documentation 1 ↗
Source-based practical guide

Extract and compare fact triples

Use FactSumm’s triple-based module to inspect facts extracted from a source article and summary.

Before you start

  • Python 3.8 or later
  • FactSumm installed with pip install factsumm

Steps

  1. Import and initialize FactSumm: from factsumm import FactSumm and factsumm = FactSumm().

  2. Set the source article and generated summary as strings.

  3. Call factsumm.extract_facts(article, summary, verbose=True) to display the extracted entities and facts for both texts.

Expected outcome

FactSumm displays source and summary facts, allowing their overlap and differences to be inspected.

Guide limitations

  • The documented module is described as triple-based and closed-scheme; no additional configuration or batch workflow is given.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
16 October 2026
Original website title
Huffon/factsumm · Official homepage title · retained in its source language
How we review sourcesSuggest a correction