Jakob Grue Simonsen

A Hierarchical Recurrent Encoder-Decoder for Generative Context-Aware Query Suggestion

Users may strive to formulate an adequate textual query for their information need. Search engines assist the users by presenting query suggestions. To preserve the original search intent, suggestions should be context-aware and account for the previous queries issued by the user. Achieving context awareness is challenging due to data sparsity. We present a novel hierarchical recurrent encoder-decoder architecture that makes possible to account for sequences of previous queries of arbitrary lengths. As a result, our suggestions are sensitive to the order of queries in the context while avoiding data sparsity. Additionally, our model can suggest for rare, or long-tail, queries. The produced suggestions are synthetic and are sampled one word at a time, using computationally cheap decoding techniques. This is in contrast to current synthetic suggestion models relying upon machine learning pipelines and hand-engineered feature sets. Results show that our model outperforms existing context-aware approaches in a next query prediction setting. In addition to query suggestion, our architecture is general enough to be used in a variety of other applications.

Is once enough?: on the extent and content of replications in human-computer interaction

A replication is an attempt to confirm an earlier studys findings. It is often claimed that research in Human-Computer Interaction (HCI) contains too few replications. To investigate this claim we examined four publication outlets (891 papers) and found 3% attempting replication of an earlier result. The replications typically confirmed earlier findings, but treated replication as a confirm/not-confirm decision, rarely analyzing effect sizes or comparing in depth to the replicated paper. When asked, most authors agreed that their studies were replications, but rarely planned them as such. Many non-replication studies could have corroborated earlier work if they had analyzed data differently or used minimal effort to collect extra data. We discuss what these results mean to HCI, including how reporting of studies could be improved and how conferences/journals may change author instructions to get more replications.

On confluence and residuals in Cauchy convergent transfinite rewriting

We show that non-collapsing orthogonal term rewriting systems do not have the transfinite Church-Rosser property in the setting of Cauchy convergence. In addition, we show that for (a transfinite version of) the Parallel Moves Lemma to hold, any definition of residual for Cauchy convergent rewriting must either part with a number of fundamental properties enjoyed by rewriting systems in the finitary and strongly convergent settings, or fail to hold for very simple rewriting systems.

Towards a Standard Testbed for Optical Music Recognition: Definitions, Metrics, and Page Images

We posit that progress in Optical Music Recognition (OMR) has been held up for years by the absence of anything resembling the standard testbeds in use in other fields that face difficult evaluation problems. One example of such a field is text information retrieval (IR), where the Text Retrieval Conference (TREC) has annually-renewed IR tasks with accompanying data sets. In music informatics, the Music Information Retrieval Exchange (MIREX), with its annual tests and meetings held during the ISMIR conference, is a close analog to TREC; but MIREX has never had an OMR track or a collection of music such a track could employ. We describe why the absence of an OMR testbed is a problem and how this problem may be mitigated. To aid in the establishment of a standard testbed, we provide (1) a set of definitions for the complexity of music notation; (2) a set of performance metrics for OMR tools that gauge score complexity and graphical quality; and (3) a small corpus of music for use as a baseline for a proper ...

Power Law Distributions in Information Retrieval

Several properties of information retrieval (IR) data, such as query frequency or document length, are widely considered to be approximately distributed as a power law. This common assumption aims to focus on specific characteristics of the empirical probability distribution of such data (e.g., its scale-free nature or its long/fat tail). This assumption, however, may not be always true. Motivated by recent work in the statistical treatment of power law claims, we investigate two research questions: (i) To what extent do power law approximations hold for term frequency, document length, query frequency, query length, citation frequency, and syntactic unigram frequency? And (ii) what is the computational cost of replacing ad hoc power law approximations with more accurate distribution fitting? We study 23 TREC and 5 non-TREC datasets and compare the fit of power laws to 15 other standard probability distributions. We find that query frequency and 5 out of 24 term frequency distributions are best approximated by a power law. All remaining properties are better approximated by the Inverse Gaussian, Generalized Extreme Value, Negative Binomial, or Yule distribution. We also find the overhead of replacing power law approximations by more informed distribution fitting to be negligible, with potential gains to IR tasks like index compression or test collection generation for IR evaluation.

An Exploration of the Relation Between Expectations and User Experience

Before using an interactive product, people form expectations about what the experience of use will be like. These expectations may affect both the use of the product and users’ attitudes toward it. This article briefly reviews existing theories of expectations to design and perform two crowdsourced experiments that investigate how expectations affect user experience measures. In the experiments, participants saw a primed or neutral review of a simple online game, played it, and rated it on various user experience measures. Results suggest that when expectations are confirmed, users tend to assimilate their ratings with their expectations; conversely, if the product quality is inconsistent with expectations, users tend to contrast their ratings with expectations and give ratings correlated with the level of disconfirmation. Results also suggest that expectation disconfirmation can be used more widely in analyses of user experience, even when the analyses are not specifically concerned with expectation disconfirmation.

On confluence of infinitary combinatory reduction systems

We prove that fully-extended, orthogonal infinitary combinatory reduction systems with finite right-hand sides are confluent modulo identification of hypercollapsing subterms. This provides the first general confluence result for infinitary higher-order rewriting.

POETS: process-oriented event-driven transaction systems

Abstract We present a high-level enterprise system architecture that closely models the domain ontology of resource and information flows in enterprises. It is: Process-oriented formal, user-definable specifications for the expected exchange of resources (money, goods, and services), notably contracts, are represented explicitly in the system state to reflect expectations on future events; Event-driven events denote relevant information about real-world transactions, specifically the transfer of resources and information between economic agents, to which the system reacts by matching against its portfolio of running processes/contracts in real time; Declarative user defined reporting functions can be formulated as declarative functions on the system state, including the representations of residual contractual obligations. We introduce the architecture and demonstrate how analyses of the standard reporting requirements for companies—the income statement and the balance sheet—can be used to drive the design of events that need registering for such reporting purposes. We then illustrate how the multi-party obligations in trade contracts (sale, purchase), including pricing and VAT payments, can be represented as formal contract expressions that can be subjected to analysis. To the best of our knowledge this is the first architecture for enterprise resource accounting that demonstrably maps high-level process and information requirements directly to executable specifications.

INFINITARY COMBINATORY REDUCTION SYSTEMS: CONFLUENCE

We study confluence in the setting of higher-order infinitary rewriting, in particular for infinitary Combinatory Reduction Systems (iCRSs). We prove that fully-extended, orthogonal iCRSs are confluent modulo identification of hypercollapsing subterms. As a corollary, we obtain that fully-extended, orthogonal iCRSs have the normal form property and the unique normal form property (with respect to reduction). We also show that, unlike the case in first-order infinitary rewriting, almost non-collapsing iCRSs are not necessarily confluent.

Sketching Shape-changing Interfaces: Exploring Vocabulary, Metaphors Use, and Affordances

Shape-changing interfaces allow designers to create user interfaces that physically change shape. However, presently, we lack studies of how such interfaces are designed, as well as what high-level strategies, such as metaphors and affordances, designers use. This paper presents an analysis of sketches made by 21 participants designing either a shape-changing radio or a shape-changing mobile phone. The results exhibit a range of interesting design elements, and the analysis points to a need to further develop or revise existing vocabularies for sketching and analyzing movement. The sketches show a prevalent use of metaphors, say, for communicating volume though big-is-on and small-is-off, as well as a lack of conventions. Furthermore, the affordances used were curiously asymmetrical compared to those offered by non-shape-changing interfaces. We conclude by offering implications on how our results can influence future research on shape-changing interfaces.