Document Chunkers

A collection of helpers to process raw documents

Overview

This library provides tools for chunking documents into manageable pieces such as paragraphs and sentences. It's particularly useful for preprocessing text data for natural language processing (NLP) tasks.

• Text Segmenters
  • Supported Models
  • Usage
    • Example: Using SaT for Sentence Segmentation
    • Example: Using WtP for Paragraph Segmentation
    • Example: Using PySBD for Sentence Segmentation
• File Formats
  • Supported File Formats
  • Usage
    • Example using MarkdownSentenceSplitter
    • Example using MarkdownParagraphSplitter
    • Example using HTMLSentenceSplitter
    • Example using HTMLParagraphSplitter
    • Example using PDFParagraphSplitter

Text Segmenters

• SaT — Segment Any Text: A Universal Approach for Robust, Efficient and Adaptable Sentence Segmentation by Markus Frohmann, Igor Sterner, Benjamin Minixhofer, Ivan Vulić and Markus Schedl (**state-of-the-art, encouraged **). - 85 languages
• WtP — Where’s the Point? Self-Supervised Multilingual Punctuation-Agnostic Sentence Segmentation by Benjamin Minixhofer, Jonas Pfeiffer and Ivan Vulić. - 85 languages
• PySBD — {P}y{SBD}: Pragmatic Sentence Boundary Disambiguation by Nipun Sadvilkar and Mark Neumann (rule-based, lightweight) - 22 languages

Usage

Example: Using SaT for Sentence Segmentation

from ovos_document_chunkers import SaTSentenceSplitter

config = {"model": "sat-3l-sm", "use_cuda": False}
splitter = SaTSentenceSplitter(config)

text = "This is a sentence. And this is another one."
sentences = splitter.chunk(text)

for sentence in sentences:
    print(sentence)

Example: Using WtP for Paragraph Segmentation

from ovos_document_chunkers import WtPParagraphSplitter

config = {"model": "wtp-bert-mini", "use_cuda": False}
splitter = WtPParagraphSplitter(config)

text = "This is a paragraph. It contains multiple sentences.\n\nThis is another paragraph."
paragraphs = splitter.chunk(text)

for paragraph in paragraphs:
    print(paragraph)

Example: Using PySBD for Sentence Segmentation

from ovos_document_chunkers import PySBDSentenceSplitter

config = {"lang": "en"}
splitter = PySBDSentenceSplitter(config)

text = "This is a sentence. This is another one!"
sentences = splitter.chunk(text)

for sentence in sentences:
    print(sentence)

File Formats

Supported File Formats

Usage

Example using MarkdownSentenceSplitter

from ovos_document_chunkers.text.markdown import MarkdownSentenceSplitter
import requests

markdown_text = requests.get("https://github.com/OpenVoiceOS/ovos-core/raw/dev/README.md").text

sentence_splitter = MarkdownSentenceSplitter()
sentences = sentence_splitter.chunk(markdown_text)

print("Sentences:")
for sentence in sentences:
    print(sentence)

Example using MarkdownParagraphSplitter

from ovos_document_chunkers.text.markdown import MarkdownParagraphSplitter
import requests

markdown_text = requests.get("https://github.com/OpenVoiceOS/ovos-core/raw/dev/README.md").text

paragraph_splitter = MarkdownParagraphSplitter()
paragraphs = paragraph_splitter.chunk(markdown_text)

print("\nParagraphs:")
for paragraph in paragraphs:
    print(paragraph)

Example using HTMLSentenceSplitter

from ovos_document_chunkers import HTMLSentenceSplitter
import requests

html_text = requests.get("https://www.gofundme.com/f/openvoiceos").text

sentence_splitter = HTMLSentenceSplitter()
sentences = sentence_splitter.chunk(html_text)

print("Sentences:")
for sentence in sentences:
    print(sentence)

Example using HTMLParagraphSplitter

from ovos_document_chunkers import HTMLParagraphSplitter
import requests

html_text = requests.get("https://www.gofundme.com/f/openvoiceos").text

paragraph_splitter = HTMLParagraphSplitter()
paragraphs = paragraph_splitter.chunk(html_text)

print("\nParagraphs:")
for paragraph in paragraphs:
    print(paragraph)

Example using PDFParagraphSplitter

from ovos_document_chunkers import PDFParagraphSplitter

pdf_path = "/path/to/your/pdf/document.pdf"

paragraph_splitter = PDFParagraphSplitter()
paragraphs = paragraph_splitter.chunk(pdf_path)

print("\nParagraphs:")
for paragraph in paragraphs:
    print(paragraph)

Credits

This work was sponsored by VisioLab, part of Royal Dutch Visio, is the test, education, and research center in the field of (innovative) assistive technology for blind and visually impaired people and professionals. We explore (new) technological developments such as Voice, VR and AI and make the knowledge and expertise we gain available to everyone.