Azure AI services
A group of Azure services, SDKs, and APIs designed to make apps more intelligent, engaging, and discoverable.
3,632 questions
I need help to understand what Azure AI tools to use to pars and query PDF files
![](data:image/svg+xml, %3Csvg xmlns='http://www.w3.org/2000/svg' height='64' class='font-weight-bold' style='font: 600 30.11764705882353px "SegoeUI", Arial' width='64'%3E%3Ccircle fill='hsl(67.2, 27%, 16%)' cx='32' cy='32' r='32' /%3E%3Ctext x='50%25' y='55%25' dominant-baseline='middle' text-anchor='middle' fill='%23FFF' %3EIK%3C/text%3E%3C/svg%3E)
Ira Korabelnikov
0
Reputation points
I need to upload PDF file 300 pages long (15MB) to pars its content and to retrieve plain text according to sections. What is best approach to do it?
I tried to use Azure Assistants - I upload the file and then prompting it but it fails to retrieve results for whole sections. I also tried to pars it with Document Inteligence and then use ChatCompletions to prompt the content of the file but get error regarding the context window despite I have 900000 TPM for this module:
"This model's maximum context length is 128000 tokens. However, your messages resulted in 212207 tokens. Please reduce the length of the messages.".
Do you have any suggestions for me?
Best Regards
Azure AI services
{count} votes
-
![](data:image/svg+xml, %3Csvg xmlns='http://www.w3.org/2000/svg' height='64' class='font-weight-bold' style='font: 600 30.11764705882353px "SegoeUI", Arial' width='64'%3E%3Ccircle fill='hsl(281.6, 88%, 37%)' cx='32' cy='32' r='32' /%3E%3Ctext x='50%25' y='55%25' dominant-baseline='middle' text-anchor='middle' fill='%23FFF' %3ESA%3C/text%3E%3C/svg%3E)
Saideep Anchuri 9,500 Reputation points Moderator2025-04-07T15:20:30.76+00:00 To parse and query a 300-page PDF file using Azure AI tools, you can utilize Azure AI Document Intelligence. This service is designed to automate document processing and can extract text from various document formats, including PDFs.
Here are some steps:
- You can use the prebuilt models available in Azure AI Document Intelligence to extract text from your PDF. The "prebuilt-read" model is specifically designed for extracting text from documents and can handle PDF files. Make sure your PDF file meets the input requirements, such as being less than 500 MB for paid tiers and ensuring that the text is legible (minimum text height of 12 pixels).
- Given the large size of your document, you may need to process it in chunks. Since the model has limitations on the number of tokens it can handle, consider breaking your PDF into smaller sections or pages before querying.
- You can integrate the Document Intelligence service using the REST API or SDKs available in languages like Python, C#, Java, and JavaScript. This will allow you to automate the process of uploading your PDF, extracting text, and querying the content based on sections.
- After extracting the text, you can implement your own logic to structure and query the content as needed
Kindly refer below link: input-requirements
Thank You.
-
Saideep Anchuri 9,500 Reputation points Moderator
2025-04-08T11:01:27.2+00:00 -
Saideep Anchuri 9,500 Reputation points Moderator
2025-04-09T10:55:55.0866667+00:00 We haven’t heard from you on the last response and was just checking back to see if you have a resolution yet.
Thank you.
-
Ira Korabelnikov 0 Reputation points
2025-04-15T16:17:23.16+00:00 Thank you for you response.
Can you suggest me way to increase accuracy of Document Intelligence. I need it to be accurate in sections extraction and structure. for example headings and subheadings levels and sizes and nested lists.
Thank you -
Saideep Anchuri 9,500 Reputation points Moderator
2025-04-15T16:35:13.7533333+00:00 To increase the accuracy of Azure AI Document Intelligence for parsing and querying PDF files, especially for extracting sections, headings, subheadings, and nested lists,.
- Use Prebuilt Models: Azure AI Document Intelligence offers various prebuilt models that are specifically designed for different document types. Utilizing these models can help in accurately extracting structured information from PDF files.
- Custom Models: If the prebuilt models do not meet your specific needs, you can create custom models tailored to your document structure. This allows for more precise extraction of headings, subheadings, and lists based on your specific formatting.
- Document Layout Analysis: Leverage the layout analysis capabilities of Document Intelligence to understand the structure of your documents better. This includes identifying paragraphs, tables, and selection marks, which can enhance the accuracy of section extraction.
- Semantic Chunking: Implement semantic chunking to break down documents into smaller, manageable sections based on headings and subheadings. This can help in maintaining the hierarchy and structure of the content.
- Fine-tuning and Training: Continuously refine and train your models with more examples of your specific document types. This iterative process can significantly improve the accuracy of extraction over time.
Kindly refer below link: document-intelligence
Thank You.
Sign in to comment
Sign in to answer