Java programmer to improve a classifier for science articles
$250-750 USD
En cours
Publié il y a plus de 13 ans
$250-750 USD
Payé lors de la livraison
Our science team needs an experienced Java programmer with experience in data mining and classification algorithms. Expert knowlede of Java SE/EE, JDBC/Hybernate is a must. Experience in bioinformatics is a definite plus. Knowledge of the Russian language is a plus.
The immediate project (for bidding is as follows):
The team has already developed an automatic text project classifier using the Space Vector Search algorithm. The algorithm looks at the database of scientific projects (title, description) and classifies the projects into the pre-defined categories.
We need to add the following functions to the classifier:
1. Establish several classification stages
At the A stage the following parameters are set: keywords, words for exclusion, search fields, fields containing settings, field containing A-stage marker.
The A stage performs the primary classification function marking the projects according to their primary features.
Example:
Keyword: aging
Words for Exclusion: children, child, family
Search Fields: title, description
Setting Fields: source = NIH
Marker field: type_res = aging_nih
At the B stage the following parameters are set: Marker field (dropdown list), keywords and words for exclusion, search fields, fields containing settings, field containing B-stage marker.
The B stage performs the classification in the project universe containing only the A stage marker and includes the project into one or several categories. The B stage marker allows to mark the projects to be able to classify the projects further with exceptions.
There needs to be a function for saving and loading the configuration for the B stage (xml format).
2. Create a Setting Analyzier
The purpose of this classifier feature is the analysis of the classification settings on search results
It loads the settings from B stage classification and displays the search results
Input: settings from B stage (dopdown list), the number of returned rows with results
Output: the table of search results with the ability to filter with set fields and sorting function.
The team is looking for a programmer to complete the project in 20 days. However, it may be possible to establish a long-term collaboration on a part-time basis.
Well, I've implemented such things. I've built automatic summarizers(my first project on freelancer.com), ontology-builders, knowledge acquisition from text systems, topic analyzers, so on and so forth. I have experience in machine learning and clusterization, I've implemented cart-s(classification and regression trees), svm, rbf-neural networks. And finally, I follow a master in Natural language processing. Hope to hear from you soon.
$250 USD en 3 jours
5,0 (2 commentaires)
2,4
2,4
7 freelances proposent en moyenne $493 USD pour ce travail
I am interested in your project for the following reasons:
1- PhD from Mai University, Japan, in Machine Learning.
2- Professor at Computer Science Faculty.
3- Required Experience are all available. [Reference Upon Request].
4- Many similar successful projects. [Reference Upon Request].
I f you are interested in interviewing me, you will receive honest answers for your questions. Payment is due upon satisfaction of work quality. I will not take the job unless you are fully satisfied that I am the right candidate.