Overview of ConceptMiner
ConceptMiner consists of the following subsystems:
FactCollector
It collects textual information according to user-defined themes in the following ways (API keys for LLM and DB services are required) :
- Competitive Products/Services
Explore the market automatically for overview of products and services based on user defined theme, and generate text that describes them using the LLM. - Sensory Experience Descriptions
Automatically explore food and beverage products on the market and collect sensory experience information for them, which LLM will then transcribe into text. - Online Articles
Collect summaries of online news articles on user-specified topics. (DB API key required) - Academic Paper Extraction
Search academic article databases and extract text from abstracts. (DB API key is required) - Patent Information Collection
Search patent information and collect patent abstract text. (DB API key is required) - Obsidian Markdown
Convert Obsidian Markdown tables to CSV.
ConceptMap-Text
We model text information stored in tabular format in CSV using our own Fuzzy Growing Batch Neural Gas (FGBNG) + Minimum Spanning Tree (MST). The model is ordered by text information only: numeric and categorical attributes can be included in the analysis, but they do not contribute to the model.
ConceptMap-Data (coming soon)
This is a derived product, a data mining tool that works with regular numerical and categorical data.
CreativeDiagram (coming soon)
Based on a network model constructed using ConceptMap-Text, we specify the space between nodes and use AI to infer new concepts corresponding to that space. We generate text such as papers, books, manuals, reports, and patent applications from the network.
T
