In this package two application examples in Saul are described. The list of the examples is listed bellow. To see more details on each example, click on its link to visit its README file.
-
Set Cover: The Set Cover problem which is a classical constraint programming problem. This example shows declarative first order constraint programming in Saul. The constraints are propositionalized and form an integer linear program (ILP) which are solved using Gurobi as our backend solver. Note that there is no training/learning involved in this example.
-
Entity-Relation Extraction: The entity-relation extraction task through which designing various training and prediction configurations are exemplified. One can see how local, global and pipeline configurations are designed, used and evaluated in Saul.
-
Spam Classification: A third example which is a binary classification task to classify text documents as either Spam or Not Spam was also created.
-
Semantic Role Labeling: a task in natural language processing consisting of the detection of the semantic arguments associated with the predicate or verb of a sentence and their classification into their specific roles.
-
Spatial Role Labeling: In Spatial Role Labeling, we try to find spatial relations and label spatial roles such as trajectors, landmarks, and spatial indicators in the sentence.
-
Part-of-Speech Tagging: Part-of-Speech Tagging is the identification of words as nouns, verbs, adjectives, adverbs, etc.
-
Twitter Sentiment Analysis: This example trains models for classifying twitter posts as positive, negative or neutral. It also includes a twitter client for real time processing of the tweets.
-
Question Type Classification: This example contains a classifications to categorize questions into different semantic classes based on the possible semantic types of the answers.
- Note: Examples are under active development.
- JDK 1.7 or Higher
- Saul
- Gurobi (required for constrained inference)