Directory Assistance Research Articles

Wi-Fi Signals Database Construction using Chebyshev Wavelets for Indoor Positioning Systems

Nowadays fast and accurate positioning of assets and people is as a crucial part of many businesses, such as, warehousing, manufacturing and logistics. Applications that offer different services based on mobile user location gaining more and more attention. Some of the most common applications include location-based advertising, directory assistance, point-to-point navigation, asset tracking, emergency and fleet management. While outdoors mostly covered by the Global Positioning System, there is no one versatile solution for indoor positioning. For the past decade Wi-Fi fingerprinting based indoor positioning systems gained a lot of attention by enterprises as an affordable and flexible solution to track their assets and resources more effectively. The concept behind Wi-Fi fingerprinting is to create signal strength database of the area prior to the actual positioning. This process is known as a calibration carried out manually and the indoor positioning system accuracy highly depends on a calibration intensity. Unfortunately, this procedure requires huge amount of time, manpower and effort, which makes extensive deployment of indoor positioning system a challenging task. approach of constructing signal strength database from a minimal number of measurements using Chebyshev wavelets approximation. The main objective of the research is to minimize the calibration workload while providing high positioning accuracy. The field tests as well as computer simulation results showed significant improvement in signal strength prediction accuracy compared to existing approximation algorithms. Furhtermore, the proposed algorithm can recover missing signal values with much smaller number of on-site measurements compared to conventional calibration algorithm.

Fact or Fraud?

Epidemic - a widespread occurrence of an infectious disease in a community at a particular time - outbreak, scourge, plague.About 1976 concerns emerged in Australia about the potential for new technologies to seriously reduce employment. The debate reached a peak in 1978 when industrial action taken by the Australian Telecommunications Employees Association (ATEA) threatened to shut down the Australian telecommunications network.From 1975 growth in calls to the directory assistance service rocketed as did the related operating costs. To maintain the quality of customer service and contain operating costs, Telecom began to deploy a nation-wide computer- based directory assistance system (DAS/C) from 1982. In 1983 an unexpected medical syndrome arose in one of Telecom's directory assistance centres.Over the next three years, the syndrome rapidly spread through other directory assistance centres, other areas of Telecom and some areas of the public service. The media, academia, legal practitioners and others attracted to the "problem" generally accepted the union view that the DAS/C system was the cause and the syndrome was labelled Repetitive Strain Injury (RSI). Medical costs and compensation claims mounted reaching $130 million in 1989.While RSI has become a well-known syndrome over the last 40 years, no outbreaks of the extent and severity experienced in Telecom appear to have been recorded over that period anywhere in the world. There still appears to be little scientific evidence of the link between the injury of the reported scale and the workplace.This is the story of the rise of the RSI phenomena in Telecom over the period 1983-86.

Fact or Fraud?

Epidemic - a widespread occurrence of an infectious disease in a community at a particular time - outbreak, scourge, plague.About 1976 concerns emerged in Australia about the potential for new technologies to seriously reduce employment. The debate reached a peak in 1978 when industrial action taken by the Australian Telecommunications Employees Association (ATEA) threatened to shut down the Australian telecommunications network.From 1975 growth in calls to the directory assistance service rocketed as did the related operating costs. To maintain the quality of customer service and contain operating costs, Telecom began to deploy a nation-wide computer- based directory assistance system (DAS/C) from 1982. In 1983 an unexpected medical syndrome arose in one of Telecom's directory assistance centres.Over the next three years, the syndrome rapidly spread through other directory assistance centres, other areas of Telecom and some areas of the public service. The media, academia, legal practitioners and others attracted to the "problem" generally accepted the union view that the DAS/C system was the cause and the syndrome was labelled Repetitive Strain Injury (RSI). Medical costs and compensation claims mounted reaching $130 million in 1989.While RSI has become a well-known syndrome over the last 40 years, no outbreaks of the extent and severity experienced in Telecom appear to have been recorded over that period anywhere in the world. There still appears to be little scientific evidence of the link between the injury of the reported scale and the workplace.This is the story of the rise of the RSI phenomena in Telecom over the period 1983-86.

List Media: The Telephone Directory and the Arranging of Names

List Media: The Telephone Directory and the Arranging of Names

Index-based incremental language model for scalable directory assistance

As the ubiquitous access to vast and remote information sources from portable devices becomes commonplace, the need from users to perform searches in keyboard-unfriendly situations grows substantially, thus triggering the increased demand of voice search sessions. This paper proposes a methodology that addresses different dimensions of scalability of mixed-initiative voice search in automatic spoken dialog systems. The strategy is based on splitting the complexity of the fully-constrained grammar (one that tightly covers the entire hypothesis space) into a fixed/low complexity phonotactic grammar followed by an index mechanism that dynamically assembles a second-pass grammar that consists of only a handful of hypotheses. The experimental analysis demonstrates different dimensions of scalability achieved by the proposed method using actual W hitepages-residential data.

Waltzing

Waltzing e.e. h. One wish. What would it be?" "Why only one? Most people get three." "We broke up," she said. "Again. For the last time." Each word was accented harshly. She blew smoke in his face. She had only been smoking for a couple of weeks, and she still had control problems. Back at his apartment, there were several burn marks on the rug, melted nylon, where she had been careless with her ashes. Here, where we worked, the floor was a safe, marred tile. "We broke up," she said. "Again. For the last time." "Come on in. You want to talk about it, or ignore it?" "Whatever works. Whatever works." At work, he was distracted even before she showed up, not that it mattered any. So late at night, it was rare for the phones to ring, and as long as the phones were quiet, he was free to do whatever he wanted. A phone company operator in a small town, the last of a dying breed. Directory assistance, maybe five calls a week. Another five calls of helping place long-distance calls. Once, an over-seas phone call. Paris. Every now and then, a panicked voice, an emergency. A choking baby. A robbery, a fire. "What does a gas leak smell like?" Nobody in their right mind used the phone in the middle of the night. It was a quiet job. He read a lot. He was being paid nearly five dollars an hour to become incredibly well read. He had a small combination t.v. and tape deck he took to work with him. Once, he talked someone out of committing suicide. Or maybe he didn't. The person agreed to flush the pills, but never called again. [End Page 102] Whether that silence meant the pills were flushed, or swallowed, he had no real way of knowing. No deaths were reported through his office. If the person did change their mind, flushed the pills, he believed they would be too embarrassed to ever call again. Instinct: never show the same person the same weakness more than once because their first instinct would be to kick it. She agreed to quit smoking that night, after she broke off with her boyfriend. If she didn't get back at him in any other way, she would simply outlive him. "I should hate men," she said. "So why am I here with you?" "I'm your best friend." "Isn't there something weird in that? Men and women just can't be friends, can they? Really? Surely not best friends." "I thought we'd been doing pretty well." "Maybe so. Maybe so." "You've been off and on with him for four years. Maybe fifteen breakups. Can you think of any fights we've had during that time?" "Only one. I decided to start smoking and you told me I was being stupid." "I was right." "Don't rub it in." "Remember that suicide I told you about? She said every time she made a mistake, her lover rubbed it in." "Rubbed what in?" "She didn't specify. I assume the mistake." "Maybe so. Maybe so." "Cancer." "Is that your sign, or your condition?" "My sign." "Lucky you." "Hey . . ." "Sorry, but I've got a call on another line." "She wanted to pick me up over the phone." He laughed. "She said I had a nice voice." [End Page 103] "You do have a nice voice." She got up from the hard metal chair, and turned on the television, not looking at it, but instead, pacing the floor. "Why do I want a cigarette? I only smoked for three weeks." "I've got a couple of joints left. In my coat. Leave the door open, and it'll suck the smoke out of the room. No one will ever know. At least they've never noticed before." "Not even the cops strolling around outside?" She shook her head. "I just want to get my lungs cleared again." She turned, and looked at the television, where Marilyn Monroe was pushing a flower pot off a ledge, barely missing Tom Ewell's head. The t.v. was snapped off...

Thanksgiving Ghosts: The Family Cookbook

Thanksgiving GhostsThe Family Cookbook Mary Ann Sternberg (bio) After a hiatus of many years, it’s my turn to host the family Thanksgiving—a perfect opportunity to introduce my new daughter-in-law and several grown-up nieces and nephews to Great Aunt Adele. About twenty-five years ago, when Thanksgiving was mine to host every year, the menu had frankly become boring. So I began to browse my cookbook shelves for inspiration and discovered a tattered, re-bound volume that had been jammed there for a decade, since my mother downsized and unloaded treasures on me. The cookbook was bound with wallpaper remnants and had no identification save a small white square adhesive patch on the front cover which had been hand lettered: “Twentieth Century.” But there was no title page—author, publisher, and publication date were a mystery. Nevertheless, I turned the frayed, yellow pages and began reading—starting with the lengthy table of contents and continuing through to “Foods for Invalids under Medical Direction” and the appendices. I was hooked when I read the chapter called “Obesity List”—foods to be avoided by “stout people,” which, unfortunately, can apply to one side of my family. When I’d finished reading, I felt like Alice fallen down the rabbit hole, into a culinary wonderland of another era. Intrigued, I phoned my mother. The cookbook had belonged to Mamie—my paternal grandmother, an excellent cook, who had lived with us when I was young. It had been written by Mamie’s sister, Aunt Adele Kahn. “You met her once when you were about two,” my mother told me. But that side of the family had scattered, and my father, the family link, had died years before. Aunt Adele’s daughter might still live in Louisville, my mother suggested provocatively. And so, in those pre-Google days, I called Louisville directory assistance for the phone number of Marion Flexner. “I beg your pardon,” demanded the strong, positive voice of the woman who answered my call. “Who exactly are you?” Cousin Marion proved to be a good sport, considering that her quiet Saturday afternoon had been invaded by someone purporting to be the granddaughter of her long-deceased Aunt Mamie who was calling to inquire about the cookbook of her own long-deceased mother. “Your cookbook,” she related with obvious pride, “was published in 1897. It was written by my mother and Mrs. C. F. Moritz.” [End Page 130] Great Aunt Adele was a published author in the late nineteenth century, before she turned twenty. Click for larger view View full resolution Great Aunt Adele was a published author in the late nineteenth century, before she turned twenty, and was “known as a clever woman, always ahead of her time.” Adele Kahn (second from the right) with the author’s grandmother, Mamie (far left), courtesy of the author. “She was known as a clever woman, always ahead of her time,” Marion said. A lifelong resident of Montgomery, Alabama, she was active in clubs and “civic matters,” in the best art and literature societies. “She even made a speech on television when she was eighty-five.” However, Marion confessed, “I never saw my mother cooking.” Recipes from the famous cookbook appeared at family dinners prepared by Molly, the family cook. “Although I’m sure my mother taught her how.” My elderly cousin had apparently inherited some of her mother’s culinary genes. She, too, was the author of cookbooks and had written numerous food articles for national magazines as well. “I really did know my way around the kitchen,” said Cousin Marion, who then lapsed into excuses for why, at age eighty-three, she was cooking much less than in years past. A month or so after our phone conversation, a package arrived with a Louisville return address. It contained a fragile, but complete copy of The Twentieth Century Cookbook and a note from Cousin Marion. “I found an extra copy of the cookbook.” [End Page 131] Click for larger view View full resolution A lifelong resident of Montgomery, Alabama, Adele Kahn Weil was active in clubs and “civic matters,” in the best art and literature societies. “She even made a...

Active learning and semi-supervised learning for speech recognition: A unified framework using the global entropy reduction maximization criterion

We propose a unified global entropy reduction maximization (GERM) framework for active learning and semi-supervised learning for speech recognition. Active learning aims to select a limited subset of utterances for transcribing from a large amount of un-transcribed utterances, while semi-supervised learning addresses the problem of selecting right transcriptions for un-transcribed utterances, so that the accuracy of the automatic speech recognition system can be maximized. We show that both the traditional confidence-based active learning and semi-supervised learning approaches can be improved by maximizing the lattice entropy reduction over the whole dataset. We introduce our criterion and framework, show how the criterion can be simplified and approximated, and describe how these approaches can be combined. We demonstrate the effectiveness of our new framework and algorithm with directory assistance data collected under the real usage scenarios and show that our GERM based active learning and semi-supervised learning algorithms consistently outperform the confidence-based counterparts by a significant margin. Using our new active learning algorithm cuts the number of utterances needed for transcribing by 50% to achieve the same recognition accuracy obtained using the confidence-based active learning approach, and by 60% compared to the random sampling approach. Using our new semi-supervised algorithm we can determine the cutoff point in determining which utterance-transcription pair to use in a principled way by demonstrating that the point it finds is very close to the achievable peak point.

Timing is everything: Methodologic issues locating and recruiting medically underserved women for abnormal mammography follow-up research

Objectives Recruiting underserved women in breast cancer research studies remains a significant challenge. We present our experience attempting to locate and recruit minority and medically underserved women identified in a Nashville, Tennessee public hospital for a mammography follow-up study. Study design The study design was a retrospective hospital-based case–control study. Methods We identified 227 women (88 African-American, 65 Caucasian, 36 other minority, 38 race undocumented in the medical record) who had undergone screening mammography and received an abnormal result during 2003–2004. Of the 227 women identified, 159 women were successfully located with implementation of a tracking protocol and more rigorous attempts to locate the women using online directory assistance and public record search engines. Women eligible for the study were invited to participate in a telephone research survey. Study completion was defined as fully finishing the telephone survey. Results An average of 4.6 telephone calls (range 1–19) and 2.7 months (range 1–490 days) were required to reach the 159 women contacted. Within three contact attempts, more cases were located than controls (61% cases vs. 49% controls, p = 0.03). African-American women cases were four times likely to be recruited than African-American controls, (OR, 4.07; 95% CI, 1.59–10.30) ( p = 0.003). After 3 months of effort, we located 67% of African-American women, 63% of Caucasian women, and 56% of other minorities. Ultimately, after a maximum of 12 attempts to contact women, 77% of African-American women and 71% of Caucasian women were eventually found. Of these, 59% of African-American women, 69% Caucasian women, and 50% other minorities were located and completed the study survey for an overall response rate of 59%, 71%, and 47% respectively. Conclusions Data collection and study recruitment efforts were more challenging in racial and ethnic minorities. Continuing attempts to contact women may increase minority group study participation but does not guarantee retention or study completion.

Search Engines as Substitutes for Traditional Information Sources? An Investigation of Media Choice

We develop a two-phased survey design—based on the uses and gratifications approach and the theory of planned behavior—to analyze competitive relations between search engines and traditional information sources. We apply the survey design in a large-scale empirical study with 14-to 66-year-old Internet users (mean age 32) to find out whether complementary or substitutional dependencies predominate between search engines and three traditional information sources—paper-based encyclopedias and yellow pages and telephone-based directory assistance. We find that search engines, compared to the traditional alternatives, are gratifying a wider spread of users' needs. Although yellow pages and directory assistance are potentially substitutable, encyclopedias serve those needs that search engines cannot (yet) fulfill. The traditional media companies face increased competition, but do not necessarily have to be in an inferior competitive position.

An Interactive Open-Vocabulary Chinese Name Input System Using Syllable Spelling and Character Description Recognition Modules for Error Correction

The open-vocabulary name recognition technique is one of the most challenging tasks in the application of automatic Chinese speech recognition technology. It can be used as the free name input method for telephony speech applications and automatic directory assistance systems. A Chinese name usually has two to three characters, each of which is pronounced as a single tonal syllable. Obviously, it is very confusing to recognize a three-syllable word from millions to billions of possible candidates. A novel interactive automatic-speech-recognition system is proposed to resolve this highly challenging task. This system was built as an open-vocabulary Chinese name recognition system using character-based approaches. Two important character-input speech-recognition modules were designed as backoff approaches in this system to complete the name input or to correct any misrecognized characters. Finite-state networks were compiled from regular grammar of syllable spellings and character descriptions for these two speech recognition modules. The possible candidate names cover more than five billions. This system has been tested publicly and proved a robust way to interact with the speaker. An 86.7% name recognition success rate was achieved by the interactive open-vocabulary Chinese name input system.

Comparing machine and human performance for caller’s directory assistance requests

To understand how to systematically construct the machine models for automating Directory Assistance (DA) that are capable of reaching the performance level of human DA operators, we conducted a number of studies over the years. This paper describes the methods used for such studies and the results of laboratory experiments. These include a series of benchmark tests configured specifically for DA related tasks to evaluate the performance of state-of-the-art and commercially-available Hidden Markov Model (HMM) based Automatic Speech Recognition (ASR) technologies. The results show that the best system achieves a 57.9% task completion rate on the city-state-recognition benchmark test. For the most frequently-requested-listing benchmark test, the best system achieves a 40% task completion rate.

Automatic speech recognition and speech variability: A review

Major progress is being recorded regularly on both the technology and exploitation of automatic speech recognition (ASR) and spoken language systems. However, there are still technological barriers to flexible solutions and user satisfaction under some circumstances. This is related to several factors, such as the sensitivity to the environment (background noise), or the weak representation of grammatical and semantic knowledge. Current research is also emphasizing deficiencies in dealing with variation naturally present in speech. For instance, the lack of robustness to foreign accents precludes the use by specific populations. Also, some applications, like directory assistance, particularly stress the core recognition technology due to the very high active vocabulary (application perplexity). There are actually many factors affecting the speech realization: regional, sociolinguistic, or related to the environment or the speaker herself. These create a wide range of variations that may not be modeled correctly (speaker, gender, speaking rate, vocal effort, regional accent, speaking style, non-stationarity, etc.), especially when resources for system training are scarce. This paper outlines current advances related to these topics.

IVR customer address acquisition method

A method of automatically acquiring and/or updating a customer address within an interactive voice response system including establishing a telephony connection with a customer and identifying a telephone number associated with the customer. A directory assistance data store external to the interactive voice response system can be accessed. A reverse directory search based upon the telephone number can be performed to retrieve an address from the directory assistance data store. The retrieved address can be stored within an interactive voice response system data store.

A split lexicon approach for improved recognition of spoken names

Recognition of spoken names is a challenging task for automatic speech recognition systems because the list of names for applications such as directory assistance tends to be in the order of several hundred thousands. This makes spoken name recognition a very high perplexity task. In this paper we propose the use of syllables as the acoustic unit for spoken name recognition based on reverse lookup schemes and show how syllables can be used to improve recognition performance and reducing the system perplexity. We present system design methodologies to address the problem of acoustic-training data sparsity encountered when using longer length units such as syllables. We illustrate our ideas first on a TIMIT based continuous speech recognition problem and then focus on the application of these ideas to spoken name recognition. Our results on the OGI spoken name corpus indicate that using syllables in place of phoneme models can help boost system accuracy significantly while helping to reduce the system complexity.

Systematic controller design to drive high-load call centers

Directory assistance services play an important role in the establishment of telephone calls, particularly business calls, today and will play an even more important role in the future. The majority of these services is provided by specialized call centers, namely DA call centers. Due to severe availability requirements and often due to very specific wishes on the part of the call center operators, the coordination of such a system is complex. This paper describes a new approach to synthesize coordination controllers using formal methods introduced by Ramadge and Wonham. The goal is on the one hand to significantly reduce the development effort, and further-through the exclusion of many potential error sources-to improve the quality and availability of the systems. The results achieved using this new approach are discussed on the basis of a prototypic implementation.

Finite-state multimodal integration and understanding

Multimodal interfaces are systems that allow input and/or output to be conveyed over multiple channels such as speech, graphics, and gesture. In addition to parsing and understanding separate utterances from different modes such as speech or gesture, multimodal interfaces also need to parse and understand composite multimodal utterances that are distributed over multiple input modes. We present an approach in which multimodal parsing and understanding are achieved using a weighted finite-state device which takes speech and gesture streams as inputs and outputs their joint interpretation. In comparison to previous approaches, this approach is significantly more efficient and provides a more general probabilistic framework for multimodal ambiguity resolution. The approach also enables tight-coupling of multimodal understanding with speech recognition. Since the finite-state approach is more lightweight in computational needs, it can be more readily deployed on a broader range of mobile platforms. We provide speech recognition results that demonstrate compensation effects of exploiting gesture information in a directory assistance and messaging task using a multimodal interface.

Automatic User-Adaptive Speaking Rate Selection

Today there are many services which provide information over the phone using a prerecorded or synthesized voice. These voices are invariant in speed. Humans giving information over the telephone, however, tend to adapt the speed of their presentation to suit the needs of the listener. This paper presents a preliminary model of this adaptation. In a corpus of simulated directory assistance dialogs the operator's speed in number-giving correlates with the speed of the user's initial response and with the user's speaking rate. Multiple regression gives a formula which predicts appropriate speaking rates, and these predictions correlate (.46) with the speeds observed in good dialogs in the corpus. It is therefore easy, at least in principle, to make systems which adapt their speed to users' needs.

Usability of dialogue design strategies for automated surname capture

Surname capture via automatic speech recognition over the telephone has many commercial applications, including automated directory assistance and travel reservation services. This paper presents a usability evaluation of three different dialogue designs for automated surname capture, within the context of a flight reservation service. The three designs explored were: a Speak Only strategy, in which callers simply say the surname; a One Stage Speak and Spell strategy in which callers speak and spell the surname in a single utterance; and a Two Stage Speak and Spell strategy in which callers speak and spell the surname in two separate dialogue stages. The methodology employed in the research provides both quantitative user attitude data and performance results for each of the strategies, based on an empirical study with a cohort of 95 participants. The results show a clear distinction between strategies. User attitude towards the dialogues that involve both speaking and spelling the name is high. User attitude towards the Speak Only strategy is significantly less positive. Task completion rates are also significantly higher in the two strategies that involve spelling the name, at around 80% compared to just over 50% in the Speak Only strategy. The data underline the importance of user testing, demonstrating the value of the evaluation methodology used, and provide encouraging results for the strategies that involve both speaking and spelling the name.

Technical Appendix B: Kansas Survey Methods

The Kansas HealthWave Evaluation Study is a multiyear study of newly enrolled children in the Kansas State Children’s Health Insurance Program (SCHIP). Families of new enrollees were interviewed near the time of program entry and again ∼12 months after enrollment. Telephone interviews were conducted with the adult in each household that was most familiar with the selected new enrollee’s health and health care experiences. The results reported in these analyses represent cross-sectional data from the new-enrollee phase of the study. Kansas’ SCHIP was implemented in January 1999 as a stand-alone supplement to the state’s preexisting Medicaid program. During the study period that began in September 2000, all children covered by HealthWave were enrolled in a single health plan offering all covered benefits except mental health services and operating on a fully capitated basis. Mental health was covered through a separately capitated carve-out. HealthWave covered otherwise-uninsured children <19 years old living in families with incomes <200% of the federal poverty level (FPL) and above the age-related Medicaid income thresholds: 150% of the FPL for children <1 year old, 133% of the …