Skip to content

Latest commit

 

History

History
103 lines (65 loc) · 3.23 KB

README.md

File metadata and controls

103 lines (65 loc) · 3.23 KB

ChemRecSys

Chemical Compounds Recommender System

This framework has implemented a Hybrid Semantic Recommender Algorithm for Chemical Compounds. It tests several recommender algorithms algorithms:

Hybrid Semantic Recommender System for Chemical Compounds

Requirements:

  • Python > 3.5
  • mySQL
  • check requirements.txt

Input:

1 - csv with a dataset with the format of <user,item,rating>, where the items are Chemical Compounds in the ChEBI ontology.

* These files may be found [here](https://drive.google.com/drive/folders/1As_BiqAAcLdUXSY2y49H-XLB87fvinVu?usp=sharing)
    
   - cheRM_20_200.csv is a sample file with 100 items
   - cheRM_20.csv is a more complet file with more than 16000 items 

* Check the creation of these files at:
    [Using Research Literature to Generate Datasets of Implicit Feedback for Recommending Scientific Items](https://ieeexplore.ieee.org/document/8924687)

2 - mySQL DB with the compounds similarities: We also need a database with the similarities between each compound in the previous csv file. This DB may be created by the framework available at https://github.com/lasigeBioTM/SemanticSimDBcreator or downloaded from here.

    - chebi_semantic_sim_cherm_20_200_dump.sql has the similarities for cheRM_20_200.csv 
    - cherm_sim_20_dump.sql has the similarities for cheRM_20.csv (more than 131 millions of similarities)

The DB have three similarities measures to be used by the recommendation algorithms: Resnik, Lin, and Jiang and Conrath (JC)

You may use the following code to dump the mySQL db.

CREATE DATABASE chebi_semantic_sim_cherm_20_200;
USE chebi_semantic_sim_cherm_20_200;

SET FOREIGN_KEY_CHECKS = 0;

DROP TABLE IF EXISTS `similarity`;
CREATE TABLE `similarity` (
  `id` INT NOT NULL AUTO_INCREMENT,
  `comp_1` INT NOT NULL,
  `comp_2` INT NOT NULL,
  `sim_resnik` FLOAT NOT NULL,
  `sim_lin` FLOAT NOT NULL,
  `sim_jc` FLOAT NOT NULL,
  PRIMARY KEY (`id`),
  INDEX sim (`comp_1`,`comp_2`) 
) ENGINE=InnoDB;


SET FOREIGN_KEY_CHECKS = 1;

exit database

mysql -u uname -p chebi_semantic_sim_cherm_20_200 < chebi_semantic_sim_cherm_20_200_dump.sql

Run:

  • The first thing to do is to change the config.ini file.

  • Second you should run the file main.py in the src folder.

  • The output will be a set of CSV files with the results for all the algorithms tested with the evaluation metrics:

    • Precision
    • Recall
    • F-Measure
    • False Positive Rate
    • Mean Reciprocal Rank
    • nDCG
    • AUC

If you wish to use a docker container, here we have an example of how to run:

  1. docker build -t "chemrec" .

  2. docker run -t -d --name chemrec_container --net=host -v /path/to/data/folder:/mlData -v /path/to/ChemRecSys:/ChemRecSys chemrec

  3. docker exec -it chemrec_container bash

  4. cd /ChemRecSys/src

  5. python main.py

The results are saved as CSV files in the folder corresponding to mlData