Sign In

Communications of the ACM

91 - 100 of 3,299 for bentley

Stranger Searching in a Strange Land: The Impact of Familiarity on Local Search

Local search entails looking for places, such as restaurants or hotels, in a geographically-constrained region. Within local search, it has been observed that an individual's familiarity with their environment (i.e. how well they know the area in a query of the form "{places} in {area}") impacts which places they are most interested in visiting. Less well-understood though is how people's information preferences differ during 1) different phases of the search process and 2) based on their level of familiarity. Through a series of surveys in the domain of dining, we explore how familiarity moderates what level of information is useful to an individual about restaurant location when choosing a place to visit. We further examine how these preferences vary between regions and phases of local search (deciding on a restaurant or determining how to go). We contribute an understanding of people's information preferences during search, building on prior research of how offline context impacts online needs.

Comparing the Reliability of Amazon Mechanical Turk and Survey Monkey to Traditional Market Research Surveys

In the product design process, it is often desirable to quickly obtain information about current user behaviors for topics that cannot be obtained through existing data or instrumentation. Perhaps we would like to understand the use of products we do not have access to or perhaps the action we would like to know about (such as using a coupon) is an action taken outside of a system that can be instrumented. Traditionally, large market research surveys would be conducted to answer these questions, but often designers need answers much faster. We present a study investigating the reliability of fast survey platforms such as Amazon Mechanical Turk and Survey Monkey as compared to larger market research studies for technology behavior research and show that results can be obtained in hours for much smaller costs with accuracy within 10% of traditional larger surveys. This demonstrates that we can rely more heavily on these platforms in the product design process and provide much faster planning iterations that are informed by actual usage data.

Automatic verification of safety and liveness for pipelined machines using WEB refinement

We show how to automatically verify that complex pipelined machine models satisfy the same safety and liveness properties as their instruction-set architecture (ISA) models by using well-founded equivalence bisimulation (WEB) refinement. We show how to reduce WEB-refinement proof obligations to formulas expressible in the decidable logic of counter arithmetic with lambda expressions and uninterpreted functions (CLU). This allows us to automate the verification of the pipelined machine models by using the UCLID decision procedure to transform CLU formulas to Boolean satisfiability problems. To relate pipelined machine states to ISA states, we use the commitment and flushing refinement maps. We evaluate our work using 17 pipelined machine models that contain various features, including deep pipelines, precise exceptions, branch prediction, interrupts, and instruction queues. Our experimental results show that the overhead of proving liveness, obtained by comparing the cost of proving both safety and liveness with the cost of only proving safety, is about 17%, but depends on the refinement map used; for example, the liveness overhead is 23% when flushing is used and is negligible when commitment is used.

The Composition and Use of Modern Mobile Phonebooks

Over the past decade, the mobile phonebook has evolved from a relatively short list of people that one calls and texts to a many-hundred person list of aggregated contacts from around the web. This is happening at a time when an increasing number of mobile applications are relying on the mobile phonebook to create one's social network in their services. Through a large-scale study of the phonebooks of 200 diverse participants, containing 65,940 contacts, we set out to understand today's mobile contact lists. Our participants reported that they did not recognize the names of 29% of their contacts and we found that the most frequently contacted five contacts represent greater than 80% of all calls and text messages with phonebook contacts. We conclude with implications for the design of mobile applications that rely on phonebook data.

Partial-match queries and file designs

Tnis paper is concernd with information retrieval based upon secondary keys; that is, keys which cannot in general uniquely identify a record, but can indicate certain attributes of the associated record. Partial-match retrieval deals with accessing and reading those records of a data base which match the user's query albeit the query is only partially specified. For example, suppose that 'the data base consists of the binary words 1010, 1110, 0011, 1101, 0010, 1111. The response to query 1**0 where * is a don't know symbol is the set of records with keys 1010 or 1110 while the response to query 1101 is the set of records with key 1101.