From 493794a7ec207a94f456b3cf5def15586d7c3f52 Mon Sep 17 00:00:00 2001 From: tmarti <2405414+tmarti@users.noreply.github.com> Date: Tue, 4 Jun 2024 09:20:51 +0200 Subject: [PATCH 1/2] Update README.md --- README.md | 16 +++++++++++++++- 1 file changed, 15 insertions(+), 1 deletion(-) diff --git a/README.md b/README.md index 22373ef..e3acca7 100644 --- a/README.md +++ b/README.md @@ -1,5 +1,7 @@ # Introducing Concept Viewer 🚀 +A visual tool to explore semantic connections between textual concepts. + ![alt text](image.png) ## Instructions @@ -15,4 +17,16 @@ yarn yarn dev ``` -Now you can navigate to http://localhost:5173/ \ No newline at end of file +Now you can navigate to http://localhost:5173/ + +## User manual + +1. Enter your OpenAI API key in the input field on the top (only used for calls to the OpenAI embedding's API) + +2. On the dropdown, select a category. This will pre-populate a list of concepts related to the category. Now, press "Represent Category". This will generate thr initial category representation. + +3. Finally, under "Your Concept" enter the concept you want explore and press "Represent Concept". You'll be able to inspect visually to which concept of thr category it lies semantically closer. + +## How it works + +When the category is represented, the sentence embeddings for its concepts are generated. Then, those embeddings are projected into the 3d-space using UMAP. The UMAP projection is then saved and reused to represent "your concept", so that the 3d space for both the category and the concept has the same "semantic coordinate system". From d4e9109ff046fc76f652641bd061c3cd61343e03 Mon Sep 17 00:00:00 2001 From: tmarti <2405414+tmarti@users.noreply.github.com> Date: Tue, 4 Jun 2024 09:21:51 +0200 Subject: [PATCH 2/2] Update README.md --- README.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/README.md b/README.md index e3acca7..424b4a3 100644 --- a/README.md +++ b/README.md @@ -19,7 +19,7 @@ yarn dev Now you can navigate to http://localhost:5173/ -## User manual +## Usage 1. Enter your OpenAI API key in the input field on the top (only used for calls to the OpenAI embedding's API)