Past Events

Two Projects in Text Data Mining and Natural Language Processing ELENA FILATOVA Department of Computer Systems Technology, New York City College of Technology, City University of New York   In this presentation I will describe two projects I am working on: Automatic Sarcasm Detection and Information Assymetries in Multilingual Wikipedia. Sarcasm detection: Humans are good at identifying sarcasm in text and speech. Can we teach a computer to identify sarcasm? Is it possible to point out the parts of the review…

The skeletal implementation pattern is a software design pattern consisting of defining an abstract class that provides a partial interface implementation. However, since Java allows only single class inheritance, if implementers decide to extend a skeletal implementation, they will not be allowed to extend any other class. Also, discovering the skeletal implementation may require a global analysis. Java 8 enhanced interfaces alleviate these problems by allowing interfaces to contain (default) method implementations, which implementers inherit. Java classes are then free to extend a different class, and a separate abstract class…

In our research, we analyze digitized images of Hematoxylin-Eosin (H&E) slides equipped with tumorous tissues from patient derived xenograft models to build our bio-inspired computation method, namely Personalized Relevance Parameterization of Spatial Randomness (PReP-SR). Applying spatial pattern analysis techniques of quadrat counts, kernel estimation and nearest neighbor functions to the images of the H&E samples, slide-specific features are extracted to examine the hypothesis that existence of dependency of nuclei positions possesses information of individual tumor characteristics. These features are then used as inputs to PReP-SR to compute tumor growth parameters for exponential-linear model. Differential evolution algorithms are developed for tumor growth parameter computations, where a candidate vector in a population consists of size selection indices for spatial evaluation and weight coefficients for spatial features and their correlations. Using leave-one-out-cross-validation method, we showed that, for a set of H&E slides from kidney cancer patient derived xenograft models, PReP-SR generates personalized model parameters with an average error rate of 13:58%. The promising results indicate that bio-inspired computation techniques may be useful to construct mathematical models with patient specific growth parameters in clinical systems.

This work introduces faceted service discovery. It uses the Programmable Web directory as its corpus of APIs and enhances the search to enable faceted search, given an OWL ontology. The ontology describes semantic features of the APIs. We have designed the API classification ontology using LexOnt, a software we have built for semi-automatic ontology creation tool. LexOnt is geared toward non-experts within a service domain who want to create a high-level ontology that describes the domain. Using well- known NLP algorithms, LexOnt generates a list of top terms and phrases from the Programmable Web corpus to enable users to find high-level features that distinguish one Programmable Web service category from another. To also aid non-experts, LexOnt relies on outside sources such as Wikipedia and Wordnet to help the user identify the important terms within a service category. Using the ontology created from LexOnt, we have created APIBrowse, a faceted search interface for APIs. The ontology, in combination with the use of the Apache Solr search platform, is used to generate a faceted search interface for APIs based on their distinguishing features. With this ontology, an API is classified and displayed underneath multiple categories and displayed within the APIBrowse interface. APIBrowse gives programmers the ability to search for APIs based on their semantic features and keywords and presents them with a filtered and more accurate set of search results.

Alexandre Passos Software Engineer, Google             In this talk we'll go over TensorFlow, an open- source cross-platform machine learning library developed by Google, and explore its new feature: eager execution. We'll go over how to use it to write dynamic models, to debug and profile models, and to learn deep learning.