Christoph Lofi

Skyline queries in crowd-enabled databases

Skyline queries are a well-established technique for database query personalization and are widely acclaimed for their intuitive query formulation mechanisms. However, when operating on incomplete datasets, skylines queries are severely hampered and often have to resort to highly error-prone heuristics. Unfortunately, incomplete datasets are a frequent phenomenon, especially when datasets are generated automatically using various information extraction or information integration approaches. Here, the recent trend of crowd-enabled databases promises a powerful solution: during query execution, some database operators can be dynamically outsourced to human workers in exchange for monetary compensation, therefore enabling the elicitation of missing values during runtime. Unfortunately, this powerful feature heavily impacts query response times and (monetary) execution costs. In this paper, we present an innovative hybrid approach combining dynamic crowd-sourcing with heuristic techniques in order to overcome current limitations. We will show that by assessing the individual risk a tuple poses with respect to the overall result quality, crowd-sourcing efforts for eliciting missing values can be narrowly focused on only those tuples that may degenerate the expected quality most strongly. This leads to an algorithm for computing skyline sets on incomplete data with maximum result quality, while optimizing crowd-sourcing costs.

Eliciting matters: controlling skyline sizes by incremental integration of user preferences

Today, result sets of skyline queries are unmanageable due to their exponential growth with the number of query predicates. In this paper we discuss the incremental re-computation of skylines based on additional information elicited from the user. Extending the traditional case of totally ordered domains, we consider preferences in their most general form as strict partial orders of attribute values. After getting an initial skyline set our basic approach aims at interactively increasing the systems information about the users wishes explicitly including indifferences. The additional knowledge then is incorporated into the preference information and constantly reduces skyline sizes. In fact, our approach even allows users to specify trade-offs between different query predicates, thus effectively decreasing the query dimensionality. We give theoretical proof for the soundness and consistence of the extended preference information and an extensive experimental evaluation of the efficiency of our approach. On average, skyline sizes can be considerably decreased in each elicitation step.

Efficient computation of trade-off skylines

When selecting alternatives from large amounts of data, trade-offs play a vital role in everyday decision making. In databases this is primarily reflected by the top-k retrieval paradigm. But recently it has been convincingly argued that it is almost impossible for users to provide meaningful scoring functions for top-k retrieval, subsequently leading to the adoption of the skyline paradigm. Here users just specify the relevant attributes in a query and all suboptimal alternatives are filtered following the Pareto semantics. Up to now the intuitive concept of compensation, however, cannot be used in skyline queries, which also contributes to the often unmanageably large result set sizes. In this paper we discuss an innovative and efficient method for computing skylines allowing the use of qualitative trade-offs. Such trade-offs compare examples from the database on a focused subset of attributes. Thus, users can provide information on how much they are willing to sacrifice to gain an improvement in some other attribute(s). Our contribution is the design of the first skyline algorithm allowing for qualitative compensation across attributes. Moreover, we also provide an novel trade-off representation structure to speed up retrieval. Indeed our experiments show efficient performance allowing for focused skyline sets in practical applications. Moreover, we show that the necessary amount of object comparisons can be sped up by an order of magnitude using our indexing techniques.

Will I Like It? Providing Product Overviews Based on Opinion Excerpts

With the growing popularity and acceptance of e-commerce platforms, users face an ever increasing burden in actually choosing the right product from the plethora of online offers. Thus, techniques for personalization and shopping assistance are in high demand by users, as well as by shopping platforms themselves. For a pleasant and successful shopping experience, users should be empowered to easily decide on which products to buy with high confidence. However, especially for entertainment goods like e.g. movies, books, or music, this task is very challenging. Unfortunately, to days approaches for dealing with this challenge (like e.g. recommender systems) suffer severe drawbacks: recommender systems are completely opaque, i.e. the recommendation is hard to justify semantically. User reviews could help users to form an opinion of recom-mended items, but with several thousand reviews available for e.g. a given popular movie, it is very challenging for users to find representative reviews. In this paper, we propose a novel technique for automatically analyzing user reviews using advanced opinion mining techniques. The results of this analysis are then used to group reviews by their semantics, i.e. by their contained opinions and point-of-views. Furthermore, the relevant paragraphs with respect to each opinion is extracted and presented to the user. These extracts can easily be digested by users to allow them a quick and diverse forming of opinion, and thus increasing their confidence in their decision, and their overall customer satisfaction.

Skyline Queries over Incomplete Data - Error Models for Focused Crowd-Sourcing

Skyline queries are a well-known technique for explorative retrieval, multi-objective optimization problems, and personalization tasks in databases. They are widely acclaimed for their intuitive query formulation mechanisms. However, when operating on incomplete datasets, skyline query processing is severely hampered and often has to resort to error-prone heuristics. Unfortunately, incomplete datasets are a frequent phenomenon due to widespread use of automated information extraction and aggregation. In this paper, we evaluate and compare various established heuristics for adapting skylines to incomplete datasets, focusing specifically on the error they impose on the skyline result. Building upon these results, we argue for improving the skyline result quality by employing crowd-enabled databases. This allows dynamic outsourcing of some database operators to human workers, therefore enabling the elicitation of missing values during runtime. Unfortunately, each crowd-sourcing operation will result in monetary and query runtime costs. Therefore, our main contribution is introducing a sophisticated error model, allowing us to specifically concentrate on those tuples that are highly likely to be error-prone, while relying on established heuristics for safer tuples. This technique of focused crowd-sourcing allows us to strike a perfect balance between costs and results quality.

iParticipate: automatic tweet generation from local government data

With the recent rise of Open Government Data, innovative technologies are required to leverage this new wealth of information. Therefore, we present a system combining several information processing techniques with micro-blogging services to demonstrate how this data can be put to use in order to increase transparency in political processes, and encourage internet users to participate in local politics. Our system uses publicly available documents from city councils which are processed automtically to generate highly informative tweets.

Mobile Product Browsing Using Bayesian Retrieval

Reacting to technological advances in the domain of mobile devices, many traditionally desktop-bound applications now are ready to make the transition into the mobile world. Especially mobile shopping applications promise a large potential for commercial. However, in order to work on the limited screen estate even of modern devices, traditional category-based browsing approaches to online shopping have to be rethought. In this paper we design an innovative approach to intuitively guide users through product databases based on Bayesian probability modeling for navigational purposes. Our navigation model is focused on feedback and inspired by content-based retrieval techniques. Moreover, we exploit new features of today’s devices like touch screens to ease interaction. Due to the novel interface-related simplicity, our system supports users in their decision process while demanding only minimal cognitive load. We outline the theoretical foundations and the design space of such a system and evaluate its retrieval effectiveness using real-world data sets. In fact, we show that using our probabilistic navigation model about 98% of all searches can be completed successfully with an average of only 3 rounds of feedback on the 4th displayed screen.

Efficiently performing consistency checks for multi-dimensional preference trade-offs

Skyline queries have recently received a lot of attention due to their intuitive query capabilities. Following the concept of Pareto optimality all dasiabestpsila database objects are returned to the user. However, this often results in unmanageable large result set sizes hampering the success of this innovative paradigm. As an effective remedy for this problem, trade-offs provide a natural concept for dealing with incomparable choices. Such trade-offs, however, are not reflected by the Pareto paradigm. Thus, incorporating them into the userspsila preference orders and adjusting skyline results accordingly needs special algorithms beyond traditional skylining. For the actual integration of trade-offs into skylines, the problem of ensuring the consistency of arbitrary trade-off sets poses a demanding challenge. Consistency is a crucial aspect when dealing with multi-dimensional trade-offs spanning over several attributes. If the consistency should be violated, cyclic preferences may occur in the result set. But such cyclic preferences cannot be resolved by information systems in a sensible way. Often, this problem is circumvented by restricting the trade-offspsila expressiveness, e.g. by altogether ignoring some classes of possibly inconsistent trade-offs. In this paper, we will present a new algorithm capable of efficiently verifying the consistency of any arbitrary set of trade-offs. After motivating its basic concepts and introducing the algorithm itself, we will also show that it exhibits superior average-case performance. The benefits of our approach promise to pave the way towards personalized and cooperative information systems.

Towards Mobile Sensor-Aware Crowdsourcing: Architecture, Opportunities and Challenges

The recent success of general purpose crowdsourcing platforms like Amazon Mechanical Turk paved the way for a plethora of crowd-enabled applications and workflows. However, the variety of tasks which can be approached via such crowdsourcing platforms is limited by constraints of the web-based interface. In this paper, we propose mobile user interface clients. Switching to mobile clients has the potential to radically change the way crowdsourcing is performed, and allows for a new breed of crowdsourcing tasks. Here, especially the ability to tap into the wealth of precision sensors embedded in modern mobile hardware is a game changer. In this paper, we will discuss opportunities and challenges resulting from such a platform, and discuss a reference architecture.

Modeling Analogies for Human-Centered Information Systems

This paper introduces a conceptual model for representing queries, statements, and knowledge in an analogy-enabled information system. Analogies are considered to be one of the core concepts of human cognition and communication, and are very efficient at conveying complex information in a natural fashion. Integrating analogies into modern information systems paves the way for future truly human-centered paradigms for interacting with data and information, and opens up a number of interesting scientific challenges, especially due to the ambiguous and often consensual nature of analogy statements. Our proposed conceptual analogy model therefore provides a unified model for representing analogies of varying complexity and type, while an additional layer of interpretation models adapts and adjusts the operational semantics for different data sources and approaches, avoiding the shortcomings of any single approach. Here, especially the Social Web promises to be a premier source of analogical knowledge due to its rich variety and subjective content, and therefore we outline first steps for harnessing this valuable information for future human-centered information systems.