Running experiments in modern computer architecture simulators can be a difficult and error-prone endeavor. Users must track many configurations, components and outputs between simulation runs. The gem5 simulator is no exception to this, requiring researchers to gather, organize, and create a significant number of components for a single simulation. In this paper, we present the gem5art framework, a tool to aid gem5 users in better structuring and running architecture simulations, and gem5 resources, a suite of resources with known compatibility with the simulator. These new additions to the gem5 project make full system simulation easier, allowing researchers to concentrate more so on their architectural innovations over setting up the simulation framework. The gem5art framework carefully logs the resources used in a gem5 simulation and places the results obtained within a database, thus enabling simple reproduction of experiments. The pre-built resources allow researchers to jump straight into running simulations rather than having to spend valuable time creating them. gem5art has been released with a permissive, open source license allowing the broader computer architecture community to contribute as workloads and workflows evolve. An archive of the data, an related materials, presented in this paper can be found at https://doi.org/10.6084/m9.figshare.14176802.

Feature selection is a core area of data mining with a recent innovation of graph-driven unsupervised feature selection for linked data. In this setting we have a dataset Y consisting of n instances each with m features and a corresponding n node graph (whose adjacency matrix is A) with an edge indicating that the two instances are similar. Existing efforts for unsupervised feature selection on attributed networks have explored either directly regenerating the links by solving for f such that f(yi,yj) ~ Ai,j or finding community structure in A and using the features in Y to predict these communities. However, graph-driven unsupervised feature selection remains an understudied area with respect to exploring more complex guidance. Here we take the novel approach of first building a block model on the graph and then using the block model for feature selection. That is, we discover FMFT ~ A and then find a subset of features S that induces another graph to preserve both F and M. We call our approach Block Model Guided Unsupervised Feature Selection (BMGUFS). Experimental results show that our method outperforms the state of the art on several real-world public datasets in finding high-quality features for clustering.

Although there have been many probabilistic or fuzzy relational database models proposed for representing and handling uncertain or imprecise information of objects in real-world applications, models combining the relevance and strength of both probability theory and fuzzy set theory appear sporadic. In this paper, we propose a new probabilistic and fuzzy relational database model where the uncertainty of a relational tuple is represented by a probability interval and the imprecision of an attribute value is represented by a fuzzy set. The mass assignment framework is employed to deal with the challenge of integration and computation of both probabilities and fuzzy sets in the same model. The fundamental concepts of the classical relational database model are extended and generalized in this new model. Then the syntax and semantics of the selection operation are formally defined accordingly.

Parallel coordinates plots (PCPs) are a well-studied technique for exploring multi-attribute datasets. In many situations, users find them a flexible method to analyze and interact with data. Unfortunately, using PCPs becomes challenging as the number of data items grows large or multiple trends within the data mix in the visualization. The resulting overdraw can obscure important features. A number of modifications to PCPs have been proposed, including using color, opacity, smooth curves, frequency, density, and animation to mitigate this problem. However, these modified PCPs tend to have their own limitations in the kinds of relationships they emphasize. We propose a new data scalable design for representing and exploring data relationships in PCPs. The approach exploits the point/line duality property of PCPs and a local linear assumption of data to extract and represent relationship summarizations. This approach simultaneously shows relationships in the data and the consistency of those relationships. Our approach supports various visualization tasks, including mixed linear and nonlinear pattern identification, noise detection, and outlier detection, all in large data. We demonstrate these tasks on multiple synthetic and real-world datasets.

Fuzzy set theory and probability theory are complementary for soft computing, in particular object-oriented systems with imprecise and uncertain object properties. However, current fuzzy object-oriented data models are mainly based on fuzzy set theory or possibility theory, and lack of a rigorous algebra for querying and managing uncertain and fuzzy object bases. In this paper, we develop an object base model that incorporates both fuzzy set values and probability degrees to handle imprecision and uncertainty. A probabilistic interpretation of relations on fuzzy sets is introduced as a formal basis to coherently unify the two types of measures into a common framework. The model accommodates both class attributes, representing declarative object properties, and class methods, representing procedural object properties. Two levels of property uncertainty are taken into account, one of which is value uncertainty of a definite property and the other is applicability uncertainty of the property itself. The syntax and semantics of the selection and other main data operations on the proposed object base model are formally defined as a full-fledged algebra.

A comprehensive product catalog is essential to the success of Product Search engines and shopping sites such as Yahoo! Shopping, Google Product Search, and Bing Shopping. Given the large number of products and the speed at which they are released to the market, keeping catalogs up-to-date becomes a challenging task, calling for the need of automated techniques. In this paper, we introduce the problem of product synthesis, a key component of catalog creation and maintenance. Given a set of offers advertised by merchants, the goal is to identify new products and add them to the catalog, together with their (structured) attributes. A fundamental challenge in product synthesis is the scale of the problem. A Product Search engine receives data from thousands of merchants about millions of products; the product taxonomy contains thousands of categories, where each category has a different schema; and merchants use representations for products that are different from the ones used in the catalog of the Product Search engine. We propose a system that provides an end-to-end solution to the product synthesis problem, and addresses issues involved in data extraction from offers, schema reconciliation, and data fusion. For the schema reconciliation component, we developed a novel and scalable technique for schema matching which leverages knowledge about previously-known instance-level associations between offers and products; and it is trained using automatically created training sets (no manually-labeled data is needed). We present an experimental evaluation using data from Bing Shopping for more than 800K offers, a thousand merchants, and 400 categories. The evaluation confirms that our approach is able to automatically generate a large number of accurate product specifications. Furthermore, the evaluation shows that our schema reconciliation component outperforms state-of-the-art schema matching techniques in terms of precision and recall.

Recent research has taken advantage of Wikipedia's multi-lingualism as a resource for cross-language information retrieval and machine translation, as well as proposed techniques for enriching its cross-language structure. The availability of documents in multiple languages also opens up new opportunities for querying structured Wikipedia content, and in particular, to enable answers that straddle different languages. As a step towards supporting such queries, in this paper, we propose a method for identifying mappings between attributes from infoboxes that come from pages in different languages. Our approach finds mappings in a completely automated fashion. Because it does not require training data, it is scalable: not only can it be used to find mappings between many language pairs, but it is also effective for languages that are under-represented and lack sufficient training samples. Another important benefit of our approach is that it does not depend on syntactic similarity between attribute names, and thus, it can be applied to language pairs that have distinct morphologies. We have performed an extensive experimental evaluation using a corpus consisting of pages in Portuguese, Vietnamese, and English. The results show that not only does our approach obtain high precision and recall, but it also outperforms state-of-the-art techniques. We also present a case study which demonstrates that the multilingual mappings we derive lead to substantial improvements in answer quality and coverage for structured queries over Wikipedia content.

There has been a substantial increase in the number of Web data sources whose contents are hidden and can only be accessed through form interfaces. To leverage this data, several applications have emerged that aim to automate and simplify the access to these data sources, from hidden-Web crawlers and meta-searchers to Web information integration systems. A requirement shared by these applications is the ability to understand these forms, so that they can automatically fill them out. In this paper, we address a key problem in form understanding: how to match elements across distinct forms. Although this problem has been studied in the literature, existing approaches have important limitations. Notably, they only handle small form collections and assume that form elements are clean and normalized, often through manual pre-processing. When a large number of forms is automatically gathered, matching form schemata presents new challenges: data heterogeneity is compounded with the Web-scale and noise introduced by automated processes. We propose PruSM, a prudent schema matching strategy the determines matches for form elements in a prudent fashion, with the goal of minimizing error propagation. A experimental evaluation of PruSM using widely available data sets shows that the approach effective and able to accurately match a large number of form schemata and without requiring any manual pre-processing.

We present DeepPeep (http://www.deeppeep.org), a new system for discovering, organizing and analyzing Web forms. DeepPeep allows users to explore the entry points to hidden-Web sites whose contents are out of reach for traditional search engines. Besides demonstrating important features of DeepPeep and describing the infrastructure we used to build the system, we will show how this infrastructure can be used to create form collections and form search engines for different domains. We also present the analysis component of DeepPeep which allows users to explore and visualize information in form repositories, helping them not only to better search and understand forms in different domains, but also to refine the form gathering process.

Wikipedia has become an important source of information which is growing very rapidly. However, the existing infrastructure for querying this information is limited and often ignores the inherent structure in the information and links across documents. In this paper, we present a new approach for querying Wikipedia content that supports a simple, yet expressive query interfaces that allow both keyword and structured queries. A unique feature of our approach is that, besides returning documents that match the queries, it also exploits relationships among documents to return richer, multi-document answers. We model Wikipedia as a graph and cast the problem of finding answers for queries as graph search. To guide the answer-search process, we propose a novel weighting scheme to identify important nodes and edges in the graph. By leveraging the structured information available in infoboxes, our approach supports queries that specify constraints over this structure, and we propose a new search algorithm to support these queries. We evaluate our approach using a representative subset of Wikipedia documents and present results which show that our approach is effective and derives high-quality answers.

We describe a machine-learning-based approach for extracting attribute labels from Web form interfaces. Having these labels is a requirement for several techniques that attempt to retrieve and integrate data that reside in online databases and that are hidden behind form interfaces, including schema matching and clustering, and hidden-Web crawlers. Whereas previous approaches to this problem have relied on heuristics and manually specified extraction rules, our technique makes use of learning classifiers to identify form labels. Our preliminary experiments show this approach is promising and has high accuracy.

In this paper we describe a new approach to extract element labels from Web form interfaces. Having these labels is a requirement for several techniques that attempt to retrieve and integrate information that is hidden behind form interfaces, such as hidden Web crawlers and metasearchers. However, given the wide variation in form layout, even within a well-defined domain, automatically extracting these labels is a challenging problem. Whereas previous approaches to this problem have relied on heuristics and manually specified extraction rules, our technique makes use of a learning classifier ensemble to identify element-label mappings; and it applies a reconciliation step which leverages the classifier-derived mappings to boost extraction accuracy. We present a detailed experimental evaluation using over three thousand Web forms. Our results show that our approach is effective: it obtains significantly higher accuracy and is more robust to variability in form layout than previous label extraction techniques.

For modelling real-world problems and constructing intelligent systems, integration of different methodologies and techniques has been the quest and focus of significant interdisciplinary research effort. The advantages of such a hybrid system are that the strengths of its partners are combined and complementary to each other's weakness. However, extended object-oriented models that combine the relevance and strength of both fuzzy set theory and probability theory appear to be sporadic. Furthermore, the soft computing paradigm needs to have real systems implemented to be useful in practice. This paper presents our development of FPDB40 as a management system for fuzzy and probabilistic object bases of the model called FPOB. The syntax and semantics of FPOB schemas, instances, and selection operation are summarized. Then the implementation of those features in FPDB40 is presented.

Although there have been many fuzzy object-oriented data models proposed, and a bit less for probabilistic ones, models combining the relevance and strength of both fuzzy set theory and probability theory appear to be sporadic. This paper introduces our extension of Eiter et al.'s probabilistic object base model with three key features: (1) uncertain and imprecise attribute values are represented as probability distributions on a set of fuzzy set values; (2) class methods with uncertain and imprecise input and output arguments are formally integrated into the new model; and (3) applicability of class properties and their inheritance can be uncertain. A probabilistic interpretation of relations on fuzzy set values is proposed for their combination with probability degrees. Then the syntax and semantics of fuzzy-probabilistic object base schemas, instances, and selection operation are presented.