Well tested & Multi-language evaluation framework for text summarization.

Testing Automatic Text Summarization

InsightLearn is an AI-driven learning and assessment platform built with Python and FastAPI. It offers smart tools such as MCQ generation, chatbot assistance, grammar correction, text summarization, and document explanation to simplify test preparation and enhance learning efficiency.

Abstract Semantic Textual Similarity (STS) measures the meaning similarity of sentences. Applications of this task include machine translation, summarization, text generation, question answering, short answer grading, semantic search, dialogue and conversational systems. We developed Support Vector Regression model with various features including the similarity scores calculated using alignment-based methods and semantic composition based methods. We have also trained sentence semantic representations with BiLSTM and Convolutional Neural Networks (CNN). The correlations between our system output the human ratings were above 0.8 in the test dataset. Introduction The goal of this task is to measure semantic textual similarity between a given pair of sentences (what they mean rather than whether they look similar syntactically). While making such an assessment is trivial for humans, constructing algorithms and computational models that mimic human level performance represents a difficult and deep natural language understanding (NLU) problem. Example 1: English: Birdie is washing itself in the water basin. English Paraphrase: The bird is bathing in the sink. Similarity Score: 5 ( The two sentences are completely equivalent, as they mean the same thing.) Example 2: English: The young lady enjoys listening to the guitar. English Paraphrase: The woman is playing the violin. Similarity Score: 1 ( The two sentences are not equivalent, but are on the same topic. ) Semantic Textual Similarity (STS) measures the degree of equivalence in the underlying semantics of paired snippets of text. STS differs from both textual entailment and paraphrase detection in that it captures gradations of meaning overlap rather than making binary classifications of particular relationships. While semantic relatedness expresses a graded semantic relationship as well, it is non-specific about the nature of the relationship with contradictory material still being a candidate for a high score (e.g., “night” and “day” are highly related but not particularly similar). The task involves producing real-valued similarity scores for sentence pairs. Performance is measured by the Pearson correlation of machine scores with human judgments.

This repository contains various models for text summarization tasks. Each model has a separate directory containing the implementation code, pretrained weights, and a Jupyter notebook for testing the model on sample input texts. Feel free to use these models for your own text summarization tasks or to experiment with them further.

Discussion Summarization is the process of condensing a text document which is a collection of discussion threads, using CBS (Cluster Based Summarization) approach in order to create a relevant summary which enlists most of the important points of the original thematic discussion, thereby providing the users, both concise and comprehensive piece of information. This outlines all the opinions which are described from multiple perspectives in a single document. This summary is completely unbiased as they present information extracted from multiple sources based on a designed algorithm, without any editorial touch or subjective human intervention. Extractive methods used here, follow the technique of selecting a subset of existing words, phrases, or sentences in the original text to form the summary. An iterative ranking algorithm is followed for clustering. The NLP (Natural Language Processing) is used to process human language data. Precisely, it is applied while working with corpora, categorizing text, analyzing linguistic structure. Thus, the quick summary is aimed at being salient, relevant and non-redundant. The proposed model is validated by testing its ability to generate optimal summary of discussions in Yahoo Answers. Results show that the proposed model is able to generate much relevant summary when compared to present summarization techniques.

Abstractive text summarization done with the help of LSTMs using encoder-decoder model which was able to achieve accuracy of 77.27% on training set and cumulative BLEU-4 score of 0.8800 on test set.

In this paper, we proposed a sequential hybrid model based on a transformer to summarize Arabic articles. We used two approaches of summarization to make our model. The First is the extractive approach which depends on the most important sentences from the articles to be the summary, so we used Deep Learning techniques specifically transformers such as AraBert to make our summary, The second is abstractive, and this approach is similar to human summarization, which means that it can use some words which have the same meaning but different from the original text. We apply this kind of summary using MT5 Arabic pre-trained transformer model. We sequentially applied these two summarization approaches to building our A3SUT hybrid model. The output of the extractive module is fed into the abstractive module. We enhanced the summary’s quality to be closer to the human summary by applying this approach. We tested our model on the ESAC dataset and evaluated the extractive summary using the Rouge score technique; we got a precision of 0.5348 and a recall of 0.5515, and an f1 score of 0.4932 and the evaluation of the abstractive model is evaluated by user satisfaction. We add some features to our summary to make it more understandable by applying the metadata generation task” data about data” and classification. By applying metadata generation, we add facilities to our summary, identification, and summary organization. Metadata provides essential contextual details, as not all summaries are self-describing. Also, classify the original text to determine the summary topic before reading. We acquire 97.5% accuracy by using Support Vector Machine (SVM) and trained it using NADA corpus.

autoTitle is a test model for Chinese text summarization

Performing abstractive summarization on dialogue-based texts poses several potential challenges to SOTA deep-learning techniques, which are tested primarily on single-author texts. I compare the performance of three SOTA pre-trained abstractive text summarization models on the TweetSum (He et al., 2020) dataset. Final project for CIS6390: Special Topics in Computing.

Learn how to build, test, and deploy a text summarization microservice with Python, FastAPI, and Docker (https://testdriven.io/courses/tdd-fastapi/)

An interactive playground for exploring Chrome's experimental built-in AI APIs. Test text generation, translation, summarization, and more directly in your browser without any server setup.

Test environment for text-summarization

Testing custom transformer for the task of text summarization

A collection of small AI experiments — text summarization, image captioning, and creative mini-projects using OpenAI models. Learn, test, and document new ideas in applied machine intelligence.

Asynchronous text summarization API built with FastAPI following Test-Driven Development (TDD) practices. Uses PostgreSQL with Tortoise ORM, Dockerized for easy deployment, and includes full CRUD endpoints, CI/CD via GitHub Actions, and optional production hosting on Heroku.

This repo essentially contains the code for the first test of text summarization software

An experimental repo for testing effective text summarization tools.

Asynchronous text summarization service API with Test-Driven Development. The API follow RESTful design principles.

Using pretrained BART model for abstract text summarization, testing the model on articles, stories and datasets.

Learn how to build, test, and deploy a text summarization microservice with Python, FastAPI, and Docker

Test-Driven Development with FastAPI and Docker, Learn how to build, test, write code/test coverage, code quality, and deploy a text summarization microservice with Python, FastAPI, and Docker!

The project is to test the foundation model framework's on device LLM capability that excels at a diverse range of text generation tasks, like summarization, entity extraction, text understanding, refinement, dialog for games, generating creative content, and more.

This project aims to learn transfer learning and fine tuning using BERT for Text Summarization case. The dataset used consists of ~190k rows, ~10k rows, ~10k rows for train, dev, and test respectively.

Document Intelligence System — A full pipeline to extract, classify, summarize, and analyze business documents (HR, Legal, Finance, Medical, Support). PDF → Text Extraction → Cleaning → NER → Document Classification → Summarization → Insights JSON. Includes standalone script, test documents.

Developed a full-stack journaling app with Java (Spring Boot) and React, featuring CRUD functionality, advanced search and filtering, and a calendar view. Designed and tested RESTful APIs with Postman and integrated AI-powered text summarization using Django and Python libraries.

This project will use natural language processing tools and algorithms to perform sentiment analysis and text summarization on yelp restaurant reviews. The evaluation will utilize training and test data and apply cross validation. The number of stars submitted in the review will define the sentiment, the target.

AI Summarizer is a PWA that uses Generative AI for fast text summarization. Built with React, TypeScript, and Express, it integrates the Gemini API to generate summaries. The app supports offline access with service workers and uses Ngrok for remote testing. Designed for students and professionals, it delivers concise summaries efficiently.

No description available

No description available