Hacker News

Article URL: https://world.hey.com/dhh/the-new-framework-13-hx370-68675e0e

Comments URL: https://news.ycombinator.com/item?id=43715036

Points: 2

# Comments: 0

Article URL: https://www.erlang-solutions.com/blog/erlang-solutions-blog-round-up/

Comments URL: https://news.ycombinator.com/item?id=43715030

Points: 2

# Comments: 0

Article URL: https://whichyr.com/

Comments URL: https://news.ycombinator.com/item?id=43715024

Points: 3

# Comments: 0

Article URL: https://www.bbc.com/news/articles/c1egdy24v7po

Comments URL: https://news.ycombinator.com/item?id=43715022

Points: 7

# Comments: 0

Article URL: https://spiraldb.com/post/so-you-want-to-use-object-storage

Comments URL: https://news.ycombinator.com/item?id=43715008

Points: 2

# Comments: 0

Article URL: https://vitalik.eth.limo/general/2025/04/14/privacy.html

Comments URL: https://news.ycombinator.com/item?id=43707999

Points: 1

# Comments: 0

Article URL: https://nitter.poast.org/paulg/status/1912147018501529937

Comments URL: https://news.ycombinator.com/item?id=43707980

Points: 5

# Comments: 0

Article URL: https://chrismanbrown.gitlab.io/66.html

Comments URL: https://news.ycombinator.com/item?id=43707976

Points: 2

# Comments: 0

Article URL: https://sherwood.news/tech/can-you-spot-the-real-ai-model-names/

Comments URL: https://news.ycombinator.com/item?id=43707965

Points: 1

# Comments: 0

Article URL: https://medicalxpress.com/news/2025-04-thalamic-nuclei-conscious-perception.html

Comments URL: https://news.ycombinator.com/item?id=43707953

Points: 1

# Comments: 0

What do you tell your growing kids in this age and time? How do you give your kids some certainty in an age that is lacking certainty?

Comments URL: https://news.ycombinator.com/item?id=43707939

Points: 2

# Comments: 4

Article URL: https://www.miamiherald.com/news/local/immigration/article304033921.html

Comments URL: https://news.ycombinator.com/item?id=43707937

Points: 2

# Comments: 0

Article URL: https://aiascendant.com/p/freaky-tales-and-johnny-canuck

Comments URL: https://news.ycombinator.com/item?id=43707936

Points: 1

# Comments: 0

Article URL: https://www.nooks.ai/blog-posts/why-all-engineers-need-to-learn-sales

Comments URL: https://news.ycombinator.com/item?id=43707931

Points: 3

# Comments: 0

Current embedding-based RAG systems primarily rely on semantic similarity. Given a document and a query, the system usually retrieves multiple sections that appear semantically relevant. However, in domain-specific applications, such as financial analysis or legal research, users often have domain-specific preferences for which parts of a document to consult first. These preferences are typically driven by experience about where answers are typically found or which sections are considered more trustworthy sources of information.

For example:

- When querying about financial performance metrics (e.g., earnings adjustments), experienced analysts typically look first at the Management’s Discussion and Analysis (MD&A) section or related financial statement footnotes.

- For questions about company risks, they usually prioritize the Risk Factors section before turning to broader disclosures.

These expert-driven navigation patterns are difficult to capture using embedding-based RAG alone. Fine-tuning embedding models to reflect such preferences is possible, but it tends to be costly and resource-intensive.

An alternative approach is to incorporate reasoning-based retrieval, which mimics how humans find information. For example, when reading a long document, a human typically starts by reviewing the table of contents to determine which sections to read first, based on the context of the query and preference. Similarly, one can build an LLM agent that analyzes the "table of contents" and then navigates through the document according to expert preferences. This can be achieved by using few-shot prompting, where the system learns from sample user preference examples provided in the prompt, allowing it to prioritize sections based on the user’s needs.

To support this paradigm, we developed an open-sourced tool called PageIndex. It can transform any long documents into an LLM-friendly "table-of-contents" tree index, which is ready for the LLM agents to navigate. With PageIndex, you can easily build RAG agents that align with user preferences and domain logic.

Would love any feedback, particularly thoughts on reasoning-based RAG or other potential applications of PageIndex.

Comments URL: https://news.ycombinator.com/item?id=43707928

Points: 7

# Comments: 1

Article URL: https://every.to/chain-of-thought/vibe-check-o3-is-out-and-it-s-great

Comments URL: https://news.ycombinator.com/item?id=43707925

Points: 2

# Comments: 1

Article URL: https://www.cnn.com/2025/04/15/science/3d-brain-map-mouse-mammal-breakthrough/index.html

Comments URL: https://news.ycombinator.com/item?id=43707919

Points: 1

# Comments: 1

Article URL: https://github.com/THUDM/GLM-4

Comments URL: https://news.ycombinator.com/item?id=43707907

Points: 1

# Comments: 0

Article URL: https://www.youtube.com/watch?v=WYPNjSoDrqw

Comments URL: https://news.ycombinator.com/item?id=43707894

Points: 1

# Comments: 0

Article URL: https://www.theregister.com/2025/04/16/trump_responds_to_nvidias_us/

Comments URL: https://news.ycombinator.com/item?id=43707891

Points: 7

# Comments: 0