Add research workflow guide

README.md

@@ -1,3 +1,95 @@
-# personal-research-knowledge-base
+# Building a Personal Research Knowledge Base
 
-Building a Personal Research Knowledge Base
+Researchers read constantly, but much of what they read is effectively lost within months. Notes are scattered across notebooks, PDF annotations, documents and emails. When a new project starts, the same papers are found and read again, and insights from previous projects are rediscovered rather than reused. A personal research knowledge base solves this problem by turning reading into a cumulative asset. This guide explains how to design one that is simple to maintain, easy to search and reliable enough to build new work on.
+
+## What a knowledge base is for
+
+A personal research knowledge base should help you do three things:
+
+1. **Find** what you have read and noted, quickly.
+2. **Trust** what you find, because each note leads back to its source.
+3. **Connect** ideas across projects, so that past reading informs new questions.
+
+Any system that achieves these three goals is good enough. Elaborate systems that are abandoned after a month are worse than simple ones maintained for years.
+
+## The building blocks
+
+### Sources
+
+Every knowledge base starts with a record of sources: papers, books, reports, datasets, talks and web pages. A reference manager is the natural home for these. Keep metadata accurate, attach the files and use consistent naming.
+
+### Notes
+
+Notes are where value accumulates. The most useful notes share a few properties:
+
+- **Atomic**: one idea per note, so it can be reused in different contexts.
+- **Written in your own words**: paraphrasing forces understanding.
+- **Linked to evidence**: each note includes a quote or precise paraphrase and a locator such as a page, table or timestamp.
+- **Dated**: so you know how current your understanding is.
+
+### Connections
+
+Connections turn a pile of notes into a knowledge base. Links between notes, shared tags and index notes that gather related ideas let you navigate from one concept to another and see patterns across sources.
+
+## A simple structure
+
+A structure that works for many researchers:
+
+- **Source notes**: one per source, summarising what it contributes, its method and its limitations, with links to the atomic notes taken from it.
+- **Concept notes**: one per idea or finding, linking to the source notes that support it.
+- **Question notes**: one per research question you care about, linking to concepts and sources relevant to it.
+- **Project notes**: one per project, listing the questions, concepts and sources used.
+
+This four-layer structure lets you enter the knowledge base from any direction: from a paper, an idea, a question or a project.
+
+## Keeping notes traceable
+
+The most important design choice is to keep every note linked to its evidence. A note that says "peer feedback improves writing quality more than teacher feedback" is dangerous without its source: which study, what population, what measure, how large an effect? Months later, you will not remember, and you may repeat the claim with more confidence than it deserves.
+
+A traceable note includes the source and a locator, and ideally a short quote. That way, when you reuse the note in a new paper, you can check the original passage in seconds.
+
+Tools that maintain these links automatically reduce the effort considerably. Research workbenches such as Yila (https://yila.ai) are built around evidence traceability: summaries and notes stay connected to the passages they came from, so revisiting the evidence behind any note is a single click rather than a search through old PDFs.
+
+## Maintaining the knowledge base
+
+A knowledge base needs light but regular maintenance:
+
+- **Process new reading weekly.** Convert highlights and margin notes into atomic notes while they are fresh.
+- **Link as you add.** When creating a note, spend a minute finding two or three related notes and linking them.
+- **Review periodically.** Once a month, browse a question note or concept note and update it with new reading.
+- **Mark superseded notes.** When later evidence changes your understanding, update the note and record what changed, rather than deleting the old view.
+
+## Searching effectively
+
+As the knowledge base grows, search becomes essential. Consistent terminology helps: decide on preferred terms for key concepts and use them in note titles. Tags should be few and meaningful. Full-text search across notes and sources covers the rest. Some tools also offer semantic search, which finds notes with related meaning even when words differ. Whatever the search method, results are only useful if each note leads back to trustworthy evidence.
+
+## Reusing knowledge in new projects
+
+When starting a new project, begin with the knowledge base rather than a fresh search. Review relevant question and concept notes, list the sources already read and identify gaps. This gives you a head start and helps you design searches that target what you do not yet know. As the project proceeds, new notes flow back into the knowledge base, making the next project easier still.
+
+## Starting small
+
+If you are starting from nothing, do not try to import years of old reading at once. Begin with the papers relevant to your current project, and add older material only when it becomes relevant again. Within a few months the knowledge base will hold the sources you actually use.
+
+## Privacy, backup and portability
+
+A personal knowledge base becomes a significant asset over years. Protect it:
+
+- back it up regularly, ideally automatically, to more than one location;
+- prefer tools that can export notes in open formats such as plain text or Markdown;
+- be careful with confidential material, such as unpublished data or interview content, and store it according to your institution's rules.
+
+## Using AI assistance with your knowledge base
+
+AI tools can help a knowledge base in several ways: suggesting links between notes, summarising clusters of notes into overview pages, and answering questions using your own notes as the source. Used this way, the AI works from material you have already read and checked, which makes its output much easier to verify. Keep the same rule as elsewhere: any AI-generated summary should cite the specific notes and source passages it relies on, and you should check those before reusing the summary in your own work.
+
+## Common mistakes
+
+- **Over-engineering.** Complex taxonomies and workflows are rarely maintained. Start simple.
+- **Collecting without processing.** Saving PDFs and highlights is not the same as building knowledge. Write notes.
+- **Losing the source.** Notes without evidence links become untrustworthy over time.
+- **Never reviewing.** A knowledge base that is only written to, never read, adds little value.
+
+## Conclusion
+
+A personal research knowledge base turns reading into a lasting asset. Keep notes atomic, written in your own words, linked to evidence and connected to each other. Maintain the system lightly but regularly, and start each new project from what you already know. Over years, the result is a growing, trustworthy map of your field that makes every new piece of research faster and better grounded.