Shady Elbassuoni

Practical extraction of disaster-relevant information from social media

During times of disasters online users generate a significant amount of data, some of which are extremely valuable for relief efforts. In this paper, we study the nature of social-media content generated during two different natural disasters. We also train a model based on conditional random fields to extract valuable information from such content. We evaluate our techniques over our two datasets through a set of carefully designed experiments. We also test our methods over a non-disaster dataset to show that our extraction model is useful for extracting information from socially-generated content in general.

Keyword search over RDF graphs

Large knowledge bases consisting of entities and relationships between them have become vital sources of information for many applications. Most of these knowledge bases adopt the Semantic-Web data model RDF as a representation model. Querying these knowledge bases is typically done using structured queries utilizing graph-pattern languages such as SPARQL. However, such structured queries require some expertise from users which limits the accessibility to such data sources. To overcome this, keyword search must be supported. In this paper, we propose a retrieval model for keyword queries over RDF graphs. Our model retrieves a set of subgraphs that match the query keywords, and ranks them based on statistical language models. We show that our retrieval model outperforms the-state-of-the-art IR and DB models for keyword search over structured data using experiments over two real-world datasets.

Robust question answering over the web of linked data

Knowledge bases and the Web of Linked Data have become important assets for search, recommendation, and analytics. Natural-language questions are a user-friendly mode of tapping this wealth of knowledge and data. However, question answering technology does not work robustly in this setting as questions have to be translated into structured queries and users have to be careful in phrasing their questions. This paper advocates a new approach that allows questions to be partially translated into relaxed queries, covering the essential but not necessarily all aspects of the users input. To compensate for the omissions, we exploit textual sources associated with entities and relational facts. Our system translates user questions into an extended form of structured SPARQL queries, with text predicates attached to triple patterns. Our solution is based on a novel optimization model, cast into an integer linear program, for joint decomposition and disambiguation of the user question. We demonstrate the quality of our methods through experiments with the QALD benchmark.

NAGA: harvesting, searching and ranking knowledge

The presence of encyclopedic Web sources, such as Wikipedia, the Internet Movie Database (IMDB), World Factbook, etc. calls for new querying techniques that are simple and yet more expressive than those provided by standard keyword-based search engines. Searching for explicit knowledge needs to consider inherent semantic structures involving entities and relationships. In this demonstration proposal, we describe a semantic search system named NAGA. NAGA operates on a knowledge graph, which contains millions of entities and relationships derived from various encyclopedic Web sources, such as the ones above. NAGAs graph-based query language is geared towards expressing queries with additional semantic information. Its scoring model is based on the principles of generative language models, and formalizes several desiderata such as confidence, informativeness and compactness of answers. We propose a demonstration of NAGA which will allow users to browse the knowledge base through a user interface, enter queries in NAGAs query language and tune the ranking parameters to test various ranking aspects.

Query relaxation for entity-relationship search

Entity-relationship-structured data is becoming more important on the Web. For example, large knowledge bases have been automatically constructed by information extraction from Wikipedia and other Web sources. Entities and relationships can be represented by subject-property-object triples in the RDF model, and can then be precisely searched by structured query languages like SPARQL. Because of their Boolean-match semantics, such queries often return too few or even no results. To improve recall, it is thus desirable to support users by automatically relaxing or reformulating queries in such a way that the intention of the original user query is preserved while returning a sufficient number of ranked results. In this paper we describe comprehensive methods to relax SPARQL-like triplepattern queries in a fully automated manner. Our framework produces a set of relaxations by means of statistical language models for structured RDF data and queries. The query processing algorithms merge the results of different relaxations into a unified result list, with ranking based on any ranking function for structured queries over RDF-data. Our experimental evaluation, with two different datasets about movies and books, shows the effectiveness of the automatically generated relaxations and the improved quality of query results based on assessments collected on the Amazon Mechanical Turk platform.

Deep answers for naturally asked questions on the web of data

We present DEANNA, a framework for natural language question answering over structured knowledge bases. Given a natural language question, DEANNA translates questions into a structured SPARQL query that can be evaluated over knowledge bases such as Yago, Dbpedia, Freebase, or other Linked Data sources. DEANNA analyzes questions and maps verbal phrases to relations and noun phrases to either individual entities or semantic classes. Importantly, it judiciously generates variables for target entities or classes to express joins between multiple triple patterns. We leverage the semantic type system for entities and use constraints in jointly mapping the constituents of the question to relations, classes, and entities. We demonstrate the capabilities and interface of DEANNA, which allows advanced users to influence the translation process and to see how the different components interact to produce the final result.

Emotion Recognition from Text Based on Automatically Generated Rules

With the growth of the Internet community, textual data has proven to be the main tool of communication in human-machine and human-human interaction. This communication is constantly evolving towards the goal of making it as human and real as possible. One way of humanizing such interaction is to provide a framework that can recognize the emotions present in the communication or the emotions of the involved users in order to enrich user experience. For example, by providing insights to users for personal preferences and automated recommendations based on their emotional state. In this work, we propose a framework for emotion classification in English sentences where emotions are treated as generalized concepts extracted from the sentences. We start by generating an intermediate emotional data representation of a given input sentence based on its syntactic and semantic structure. We then generalize this representation using various ontologies such as Word Net and Concept Net, which results in an emotion seed that we call an emotion recognition rule (ERR). Finally, we use a suite of classifiers to compare the generated ERR with a set of reference ERRs extracted from a training set in a similar fashion. The used classifiers are k-nearest neighbors (KNN) with handcrafted similarity measure, Point Mutual Information (PMI), and PMI with Information Retrieval (PMI-IR). When applied on different datasets, the proposed approach significantly outperformed the existing state-of-the art machine learning and rule-based classifiers with an average F-Score of 84%.

CATE: context-aware timeline for entity illustration

Wikipedia has become one of the most authoritative information sources on the Web. Each article in Wikipedia provides a portrait of a certain entity. However, such a portrait is far from complete. An informative portrait of an entity should also reveal the context the entity belongs to. For example, for a person, major historical, political and cultural events that coincide with her life are important and should be included in that persons portrait. Similarly, the persons interactions with other people are also important. All this information should be summarized and presented in an appealing and interactive visual interface that enables users to quickly scan the entitys portrait. We demonstrate CATE which is a system that utilizes Wikipedia to create a portrait of a given entity of interest. We provide a visualization tool that summarizes the important events related to the entity. The novelty of our approach lies in seeing the portrait of an entity in a broader context, synchronous with its time.

Task-aware search personalization

Search personalization has been pursued in many ways, in order to provide better result rankings and better overall search experience to individual users [5]. However, blindly applying personalization to all user queries, for example, by a background model derived from the users long-term query-and-click history, is not always appropriate for aiding the user in accomplishing her actual task. User interests change over time, a user sometimes works on very different categories of tasks within a short timespan, and history-based personalization may impede a users desire of discovering new topics. In this paper we propose a personalization framework that is selective in a twofold sense. First, it selectively employs personalization techniques for queries that are expected to benefit from prior history information, while refraining from undue actions otherwise. Second, we introduce the notion of tasks representing different granularity levels of a user profile, ranging from very specific search goals to broad topics, and base our reasoning selectively on query-relevant user tasks. These considerations are cast into a statistical language model for tasks, queries, and documents, supporting both judicious query expansion and result re-ranking. The effectiveness of our method is demonstrated by an empirical user study.

S3K: seeking statement-supporting top-K witnesses

Traditional information retrieval techniques based on keyword search help to identify a ranked set of relevant documents, which often contains many documents in the top ranks that do not meet the users intention. By considering the semantics of the keywords and their relationships, both precision and recall can be improved. Using an ontology and mapping keywords to entities/concepts and identifying the relationship between them that the user is interested in, allows for retrieving documents that actually meet the users intention. In this paper, we present a framework that enables semantic-aware document retrieval. User queries are mapped to semantic statements based on entities and their relationships. The framework searches for documents expressing these statements in different variations, e.g., synonymous names for entities or different textual expressions for relations between them. The size of potential result sets makes ranking documents according to their relevance to the user an essential component of such a system. The ranking model proposed in this paper is based on statistical language-models and considers aspects such as the authority of a document and the confidence in the textual pattern representing the queried information.