kitconc 3.4.3

A toolkit for Corpus Linguistics Analysis

Kitconc 3.4.3

Kitconc is a package for Corpus Linguistics and text analysis with Python. It contains, among other things, tools for creating:

• Corpora;

• Frequency wordlists;

• Keywords (Log-Likelihood, Chi-Square, TF-IDF);

• Concordance lines;

• Collocates;

• N-gram lists;

• Dispersion plots;

• Excel data files;

• Semantic search with sentence embeddings.

The package is built on top of platforms and packages for scientific research: numpy, pandas, NLTK, XlsxWriter and matplotlib.

Requirements

Kitconc requires Python 3.10 or later.

Package dependencies (pip install kitconc):

numpy>=1.26.4,<2.0.0 pandas>=2.2.0,<3.0.0 matplotlib>=3.7.0,<4.0.0 xlsxwriter>=3.2.3,<4.0.0 ttkbootstrap>=1.12.0,<2.0.0 pillow>=11.2.0,<12.0.0 requests>=2.31.0,<3.0.0 nltk>=3.9.1,<4.0.0 chardet>=5.2.0,<6.0.0 pypdf>=4.0.0,<7.0.0 cryptography>=3.1,<47.0.0 mcp>=1.0.0,<2.0.0 setuptools>=70.0.0

Additional dependencies listed in requirements.txt (full local environment):

torch>=2.6,<2.10 (CPU wheels via –extra-index-url https://download.pytorch.org/whl/cpu) transformers>=4.45,<6.0.0 sentence-transformers>=3.0,<6.0.0 sqlite-vec>=0.1.7,<1.0.0 fastapi>=0.110,<1.0.0 uvicorn[standard]>=0.27,<1.0.0 python-dotenv>=1.0.0,<2.0.0

Installation

pip install kitconc

Kitconc App (graphical interface)

kitconc-app

Agent Layer (internal actions)

Kitconc now includes an internal action layer for agent/tool orchestration:

• kitconc.agent.actions.KitconcActions

• Full parity with shell commands from kit_cmd.py (do_*)

• Typed schemas in kitconc.agent.schemas

• Contract documentation in kitconc/agent/CONTRACT.md

• Semantic retrieval action: semantic_search(…)

Basic usage:

from kitconc.agent import KitconcActions actions = KitconcActions(“kitconc_workspace”) actions.create(“ads”, “kitconc_corpora/ads”, “english”) actions.use(“ads”) rows = actions.keywords(limit=10)

MCP Server (for agent integrations)

kitconc-mcp –transport stdio

For HTTP clients (recommended):

kitconc-mcp –transport streamable-http –host 127.0.0.1 –port 8001

or (legacy SSE):

kitconc-mcp –transport sse –host 127.0.0.1 –port 8001

Includes semantic retrieval tool: semantic_search (query, top_k, db_path, model_name)

MCP runtime is included in package dependencies (pip install kitconc is enough).

What’s new in 3.2.0

• Tkinter launcher command – start GUI with kitconc-app

• Agent action layer – kitconc.agent.actions.KitconcActions with command parity from kit_cmd.py

• Typed schemas – available in kitconc.agent.schemas

• MCP server entrypoint – run with kitconc-mcp

• Semantic search MCP tool – semantic_search for sqlite-vec indexes

• Embedding index hardening – transactional writes and thread-safe SQLite access

• Progress flag rename – use verbose=True (replacing show_progress=True)

What’s new in 3.1.0

• TF-IDF keywords – third keyword extraction method alongside Log-Likelihood and Chi-Square

• Keyword filters – ignore numbers, ignore words with strange characters, minimum word length

• PDF support – add PDF files directly to a corpus

• Embeddings module – semantic search with sentence-transformers and SQLite vector storage

• Dialog improvements – dialog boxes now center correctly in fullscreen and large-window mode

Language resources

Kitconc comes with built-in language resources for Portuguese and English corpora. It also provides functions for adding your own language resources.

Usage example

See how easy it is to use Kitconc:

https://ilexis.net.br/kitconc