About Me
I am an Assistant Professor at Computer Science Department, Faculty of Science, Minia University, Egypt.
Please check my email: CV1 and CV2
You can send me an email at osmaneg200@gmail.com.
------------------------------------------------------------------------------------------------------------------------------------------------------------------------
ES-Rank, EGS-Rank, SAS-Rank, VNS and GWO
Osman Ali Sadek Ibrahim (osmaneg200@gmail.com)
An Efficient Tool for Preference Ranking
Optimizing data features help to facilitate the instance selection and analysis the dataset for ranking problems. Sometimes research problem such as ranking dataset instances in Medical Diagnosis, Search Engine and Information Retrieval require ranking models for the data instances based on the importance of the features on the datasets. These tools are efficient metaheuristic tools that produce ranking models based on the training & validation datasets and evaluating the produced model on the test datasets for predictive results.
Download: (ES-Rank), (EGS-Rank), (SAS-Rank), (GVN-Rank) and (Multi-ES-Rank) (under development for publishing more research outcomes).
Download example files:
These tools can be used for both Preference Learning To Rank. These tools use 7 fitness evaluation metrics {Mean Average Precision (MAP), Discounted Cumulative Gain (DCG@K), Normalized Discounted Cumulative Gain (NDCG@K), Precision (P@K), Reciprocal Rank (RR@K), Best (BEST@K) and Expected Reciprocal Rank (ERR@K)} with variable parameters. The default fitness evaluation function is MAP. The parameter k is the number of top-k retrieved instance from the ranking model by ES-Rank that requires measuring for them the evaluation metric value.
The training set, validation set and test set format can be in one of these two formats:
1) Label FeatureID:FeatureValue FeatureID:FeatureValue
OR
2) Label QueryID:QueryNo. FeatureID:FeatureValue FeatureID:FeatureValue
Example for training and testing set are:
1)
7 1:0.3 2:1
10 1:0.6 2:1.2
11 1:3 2:8
2)
7 qid:100 1:0.3 2:1
10 qid:100 1:0.6 2:1.2
11 qid:101 1:3 2:8
To transfer space, tab or comma delimited data files and convert them into the format required (the above format), you can use this EXE file (CSVtoLTR) in cmd command line.
Example for conversion: c:\>CSVtoLTR.exe (then ENTER and follow the instructions).
To use the package in Windows command line or Linux Terminal, you can use the following command:
java -jar IESRank.jar (Then ENTER to check the tool instruction and example).
For reference the tool, please use reference [1] at the moment.
(Note: the tool under development for publishing two improving outcomes).
References:
[1] Ismail, Walaa N., Osman Ali Sadek Ibrahim, Hessah A. Alsalamah, and Ebtesam Mohamed, Multiobjective Learning to Rank Based on the (1 + 1) Evolutionary Strategy: An Evaluation of Three Novel Pareto Optimal Methods, Electronics 12, no. 17: 3724, 2023. https://doi.org/10.3390/electronics12173724
[2] Osman A. S. Ibrahim and Eman M. G. Younis, Combining variable neighborhood with gradient ascent for learning to rank problem, Neural Computing and Applications, Springer Publisher, 2023 (pdf).
[3] Osman A. S. Ibrahim and Eman M. G. Younis, Hybrid Online-Offline Learning to Rank Using Simulated Annealing Strategy Based on Dependent Click Model, Knowledge and Information Systems Journal, Springer Publisher, 2022 (pdf).
[4] Osman A. S. Ibrahim and Dario Landa-Silva, ES-Rank: Evolution Strategy Learning to Rank Approach, ACM Symposium on Applied Computing (SAC 2017), Marrakech, Morocco, April 03-07, 2017, ISBN: 978-1-4503-4486-9/17/04 (pdf).
[5] Osman A. S. Ibrahim, Evolutionary Algorithms and Machine Learning Techniques for Information Retrieval, PhD Thesis, ASAP Research Group, School of Computer Science, The University of Nottingham, Jubilee Campus, UK (pdf).
[6] Osman A. S. Ibrahim and Dario Landa-Silva, An Evolutionary Strategy with Machine Learning for Learning to Rank in Information Retrieval, Accepted in Soft Computing Journal.DOI: 10.1007/s00500-017-2988-6 (pdf).