Short text categorization is an important task due to the rapid growth of online available short texts in various domains such as web search snippets, etc. Most of the traditional methods suffer from sparsity and shortness of the text. Moreover, supervised learning methods require a significant amount of training data and manually labeling such data can be very time-consuming and costly. In this study, we propose a novel probabilistic model for Knowledge-Based Short Text Categorization (KBSTC), which does not require any labeled training data to classify a short text. This is achieved by leveraging entities and categories from large knowledge bases, which are further embedded into a common vector space, for which we propose a new entity and category embedding model. Given a short text, its category (e.g. Business , Sports , etc.) can then be derived based on the entities mentioned in the text by exploiting semantic similarity between entities and categories. To validate the effectiveness of the proposed method, we conducted experiments on two real-world datasets, i.e., AG News and Google Snippets. The experimental results show that our approach significantly outperforms the classification approaches which do not require any labeled data, while it comes close to the results of the supervised approaches.

Extracting knowledge from Wikipedia has attracted much attention in recent ten years. One of the most valuable kinds of knowledge is type information, which refers to the axioms stating that an instance is of a certain type. Current approaches for inferring the types of instances from Wikipedia mainly rely on some language-specific rules. Since these rules cannot catch the semantic associations between instances and classes (i.e. candidate types), it may lead to mistakes and omissions in the process of type inference. The authors propose a new approach leveraging attributes to perform language-independent type inference of the instances from Wikipedia. The proposed approach is applied to the whole English and Chinese Wikipedia, which results in the first version of MulType (Multilingual Type Information), a knowledge base describing the types of instances from multilingual Wikipedia. Experimental results show that not only the proposed approach outperforms the state-of-the-art comparison methods, but also MulType contains lots of new and high-quality type information.

In recent years, the amount of entities in large knowledge bases available on the Web has been increasing rapidly, making it possible to propose new ways of intelligent information access. Within the context of globalization, there is a clear need for techniques and systems that can enable multilingual and cross-lingual information access. In this paper, we present XKnowSearch!, a novel entity-based system for multilingual and cross-lingual information retrieval, which supports keyword search and also allows users to influence the search process according to their search intents. By leveraging the multilingual knowledge base on the Web, keyword queries and documents can be represented in their semantic forms, which can facilitate query disambiguation and expansion, and can also overcome the language barrier between queries and documents in different languages.

In recent years, the amount of entities in large knowledge bases has been increasing rapidly. Such entities can help to bridge unstructured text with structured knowledge and thus be beneficial for many entity-centric applications. The key issue is to link entity mentions in text with entities in knowledge bases, where the main challenge lies in mention ambiguity. Many methods have been proposed to tackle this problem. However, most of the methods assume certain characteristics of the input mentions and documents, e.g., only named entities are considered. In this paper, we propose a context-aware approach to collective entity disambiguation of the input mentions in text with different characteristics in a consistent manner. We extensively evaluate the performance of our approach over 9 datasets and compare it with 14 state-of-the-art methods. Experimental results show that our approach outperforms the existing methods in most cases.

Xavier Carreras, Lluís Padró, Lei Zhang, Achim Rettinger, Zhixing Li, Esteban García-Cuesta, Željko Agić, Božo Bekavac, Blaz Fortuna, Tadej Štajner. Proceedings of the Demonstrations at the 14th Conference of the European Chapter of the Association for Computational Linguistics. 2014.

Within the context of globalization, multilinguality and cross-linguality for information access have emerged as issues of major interest. In order to achieve the goal that users from all countries have access to the same information, there is an impending need for systems that can help in overcoming language barriers by facilitating multilingual and cross-lingual access to data. In this paper, we demonstrate such a toolkit, which supports both service-oriented and user-oriented interfaces for semantically annotating, analyzing and comparing multilingual texts across the boundaries of languages. We conducted an extensive user study that shows that our toolkit allows users to solve cross-lingual entity tracking and article matching tasks more efficiently and with higher accuracy compared to the baseline approach. 1

The ever-increasing quantities of structured knowledge on the Web and the impending need of multilinguality and cross-linguality for information access pose new challenges but at the same time open up new opportunities for knowledge extraction research. In this regard, cross-lingual semantic annotation has emerged as a topic of major interest and it is essential to build tools that can link words and phrases in unstructured text in one language to resources in structured knowledge bases in any other language. In this paper, we demonstrate X-LiSA, an infrastructure for cross-lingual semantic annotation, which supports both service-oriented and user-oriented interfaces for annotating text documents and web pages in different languages using resources from Wikipedia and Linked Open Data (LOD).

Keyword search is an intuitive paradigm for searching linked data sources on the web. We propose to route keywords only to relevant sources to reduce the high cost of processing keyword search queries over all sources. We propose a novel method for computing top-k routing plans based on their potentials to contain results for a given keyword query. We employ a keyword-element relationship summary that compactly represents relationships between keywords and the data elements mentioning them. A multilevel scoring mechanism is proposed for computing the relevance of routing plans based on scores at the level of keywords, data elements, element sets, and subgraphs that connect these elements. Experiments carried out using 150 publicly available sources on the web showed that valid plans (precision@1 of 0.92) that are highly relevant (mean reciprocal rank of 0.89) can be computed in 1 second on average on a single PC. Further, we show routing greatly helps to improve the performance of keyword search, without compromising its result quality.

The problem of rewriting keyword search queries on graph data has been studied recently, where the main goal is to clean user queries by rewriting keywords as valid tokens appearing in the data and grouping them into meaningful segments. The main solution to this problem employs heuristics for ranking query rewrites and a dynamic programming algorithm for computing them. Based on a broader set of queries defined by an existing benchmark, we show that the use of these heuristics does not yield good results. We propose a novel probabilistic framework, which enables the optimality of a query rewrite to be estimated in a more principled way. We show that our approach outperforms existing work in terms of effectiveness and efficiency of query rewriting. More importantly, we provide the first results indicating query rewriting can indeed improve overall keyword search runtime performance and result quality.