Varish Mulwad

Semantic Message Passing for Generating Linked Data from Tables

We describe work on automatically inferring the intended meaning of tables and representing it as RDF linked data, making it available for improving search, interoperability and integration. We present implementation details of a joint inference module that uses knowledge from the linked open data (LOD) cloud to jointly infer the semantics of column headers, table cell values (e.g., strings and numbers) and relations between columns. We also implement a novel Semantic Message Passing algorithm which uses LOD knowledge to improve existing message passing schemes. We evaluate our implemented techniques on tables from the Web and Wikipedia.

Extracting Information about Security Vulnerabilities from Web Text

The Web is an important source of information about computer security threats, vulnerabilities and cyber attacks. We present initial work on developing a framework to detect and extract information about vulnerabilities and attacks from Web text. Our prototype system uses Wikitology, a general purpose knowledge base derived from Wikipedia, to extract concepts that describe specific vulnerabilities and attacks, map them to related concepts from DBpedia and generate machine understandable assertions. Such a framework will be useful in adding structure to already existing vulnerability descriptions as well as detecting new ones. We evaluate our approach against vulnerability descriptions from the National Vulnerability Database. Our results suggest that it can be useful in monitoring streams of text from social media or chat rooms to identify potential new attacks and vulnerabilities or to collect data on the spread and volume of existing ones.

A domain independent framework for extracting linked semantic data from tables

Vast amounts of information is encoded in tables found in documents, on the Web, and in spreadsheets or databases. Integrating or searching over this information benefits from understanding its intended meaning and making it explicit in a semantic representation language like RDF. Most current approaches to generating Semantic Web representations from tables requires human input to create schemas and often results in graphs that do not follow best practices for linked data. Evidence for a tables meaning can be found in its column headers, cell values, implicit relations between columns, caption and surrounding text but also requires general and domain-specific background knowledge. Approaches that work well for one domain, may not necessarily work well for others. We describe a domain independent framework for interpreting the intended meaning of tables and representing it as Linked Data. At the core of the framework are techniques grounded in graphical models and probabilistic reasoning to infer meaning associated with a table. Using background knowledge from resources in the Linked Open Data cloud, we jointly infer the semantics of column headers, table cell values (e.g., strings and numbers) and relations between columns and represent the inferred meaning as graph of RDF triples. A tables meaning is thus captured by mapping columns to classes in an appropriate ontology, linking cell values to literal constants, implied measurements, or entities in the linked data cloud (existing or new) and discovering or and identifying relations between columns.

CyberTwitter: using Twitter to generate alerts for cybersecurity threats and vulnerabilities

In order to secure vital personal and organizational system we require timely intelligence on cybersecurity threats and vulnerabilities. Intelligence about these threats is generally available in both overt and covert sources like the National Vulnerability Database, CERT alerts, blog posts, social media, and dark web resources. Intelligence updates about cybersecurity can be viewed as temporal events that a security analyst must keep up with so as to secure a computer system. We describe CyberTwitter, a system to discover and analyze cybersecurity intelligence on Twitter and serve as a OSINT (Open-source intelligence) source. We analyze real time information updates, in form of tweets, to extract intelligence about various possible threats. We use the Semantic Web RDF to represent the intelligence gathered and SWRL rules to reason over extracted intelligence to issue alerts for security analysts.

Interpreting medical tables as linked data for generating meta-analysis reports

Evidence-based medicine is the application of current medical evidence to patient care and typically uses quantitative data from research studies. It is increasingly driven by data on the efficacy of drug dosages and the correlations between various medical factors that are assembled and integrated through meta-analyses (i.e., systematic reviews) of data in tables from publications and clinical trial studies. We describe a important component of a system to automatically produce evidence reports that performs two key functions: (i) understanding the meaning of data in medical tables and (ii) identifying and retrieving relevant tables given a input query. We present modifications to our existing framework for inferring the semantics of tables and an ontology developed to model and represent medical tables in RDF. Representing medical tables as RDF makes it easier for the automatic extraction, integration and reuse of data from multiple studies, which is essential for generating meta-analyses reports. We show how relevant tables can be identified by querying over their RDF representations and describe two evaluation experiments: one on mapping medical tables to linked data and another on identifying tables relevant to a retrieval query.

Industrial Knowledge Graphs Workshop 2017: co-located with the 9th International ACM Web Science Conference 2017 (Preface; part of Content)

This document provides a summary of the Industrial Knowledge Graph Workshop co-located with the ACM Web Science 2017 conference held on June 25, 2017, in Troy, NY, USA

Interactive online learning for clinical entity recognition

Named entity recognition and entity linking are core natural language processing components that are predominantly solved by supervised machine learning approaches. Such supervised machine learning approaches require manual annotation of training data that can be expensive to compile. The applicability of supervised, machine learning-based entity recognition and linking components in real-world applications can be hindered by the limited availability of training data. In this paper, we propose a novel approach that uses ontologies as a basis for entity recognition and linking, and captures context of neighboring tokens of the entities of interest with vectors based on syntactic and semantic features. Our approach takes user feedback so that the vector-based model can be continuously updated in an online setting. Here we demonstrate our approach in a healthcare context, using it to recognize body part and imaging modality entities within clinical documents, and map these entities to the right concepts in the RadLex and NCIT medical ontologies. Our current evaluation shows promising results on a small set of clinical documents with a precision and recall of 0.841 and 0.966. The evaluation also demonstrates that our approach is capable of continuous performance improvement with increasing size of examples. We believe that our human-in-the-loop, online learning approach to entity recognition and linking shows promise that it is suitable for real-world applications.

Semantic Interpretation of Structured Log Files

Data from computer log files record traces of events involving user activity, applications, system software and network traffic. Logs are usually intended for diagnostic and debugging purposes, but their data can be extremely useful in system audits and forensic investigations. Logs created by intrusion detection systems, Web servers, antivirus and anti-malware systems, firewalls and network devices have information that can reconstruct the activities of malware or a malicious agent, help plan for remediation and prevent attacks by revealing probes or intrusions before damage has been done. While existing tools like Splunk can help analyze logs with known schemas, understanding log whose format is unfamiliar or associated with new device or custom application can be challenging. We describe a framework for analyzing logs and automatically generating a semantic description of their schema and content in RDF. The framework begins by normalizing the log into columns and rows using regular expression-based and dictionary-based classifiers. Leveraging our existing work on inferring the semantics of tables, we associate semantic types with columns and, when possible, map them to concepts in general knowledge-bases (e.g. DBpedia) and domain specific ones (e.g., Unified Cybersecurity Ontology). We link cell values to known type instances (e.g., an IP address) and suggest relationships between columns. Converting large and verbose log files into such semantic representations reveals their meaning and supports search, integration and reasoning over the data.

DC proposal: graphical models and probabilistic reasoning for generating linked data from tables

Vast amounts of information is encoded in tables found in documents, on the Web, and in spreadsheets or databases. Integrating or searching over this information benefits from understanding its intended meaning and making it explicit in a semantic representation language like RDF. Most current approaches to generating Semantic Web representations from tables requires human input to create schemas and often results in graphs that do not follow best practices for linked data. Evidence for a tables meaning can be found in its column headers, cell values, implicit relations between columns, caption and surrounding text but also requires general and domain-specific background knowledge. We describe techniques grounded in graphical models and probabilistic reasoning to infer meaning associated with a table. Using background knowledge from the Linked Open Data cloud, we jointly infer the semantics of column headers, table cell values (e.g., strings and numbers) and relations between columns and represent the inferred meaning as graph of RDF triples. A tables meaning is thus captured by mapping columns to classes in an appropriate ontology, linking cell values to literal constants, implied measurements, or entities in the linked data cloud (existing or new) and discovering or and identifying relations between columns.