京都大学 / 0048 / 新制・課程博士 / 博士(情報学) / 甲第20026号 / 情博第621号 / 新制||情||108(附属図書館) / 33122 / 京都大学大学院情報学研究科社会情報学専攻 / (主査)教授 田中 克己, 教授 吉川 正俊, 教授 黒橋 禎夫 / 学位規則第4条第1項該当 / Doctor of Informatics / Kyoto University / DFAM

<p> </p> <p>A search and rescue (SAR) operation requires a rapid, accurate, and effective response to provide the missing person the best chance of being rescued. Personnel from the local area are likely to be closest to the location of the missing person, be familiar with the area, but they may not be adequately trained, experienced, or equipped with the best tools to effectively locate, identify, and retrieve the missing person. Thus, most SAR operations rely on a mix of trained personnel and volunteers. Among the trained personnel, there is a wide variance in proficiency, experience, and access to technology, leading to some emergency response agencies being better prepared than others. Volunteers, on the other hand, could be very helpful, but are largely untrained and inexperienced, reducing their inherent likelihood of success. The primary challenge to successful SAR operations is the lack of consistently trained, adequately equipped, and diversely experienced personnel. Despite the lack of desired resources, SAR operations must be completed rapidly and emergency responders often turn to volunteers. In response to this challenge, the use of unmanned aerial systems, UAS, in small volunteer teams was proposed. Available, off-the-shelf UAS technology can be used to simplify training with the help of affordable advanced technology, and thereby enable rapid, accurate, and effective SAR operations.  </p> <p>The following research was executed in the form of three independent, but related, studies. The first study focused on the efficiency of a UAS-equipped SAR operation; the second study focused on the accuracy of a UAS and image analysis software-enabled SAR operation; and the third study tested the ability of novice volunteers to learn and apply the new technologies (UAS plus image analysis) efficiently and effectively. The goal of these studies was to determine whether affordable commercial, off-the-shelf technologies could be used to enhance the efficiency and effectivity of SAR operations. The experimental methodology used specifically designed simulations of SAR operational scenarios. Two operational tactics were tested: (a) Equip the SAR team with UAS and (b) equip the SAR team with UAS and image analysis software. The specific scenarios selected were similar in complexity, but different enough to minimize the transfer of learning from the first study to the second study.  Finally, the reference times for manual SAR operations were compared against UAS and computerized image analysis software-assisted methods. The results of the proposed studies determined whether off-the-shelf UAS and image analysis technologies could be used to enable rapid, accurate, and effective SAR operations. </p>

Using search to navigate an application could sometimes be a more efficient way than browsing. This does however require the implementation of the search to be satisfactory for a user not to abandon it. This thesis focused on improving a keyword-only solution by studying how users construct navigational queries and implementing a new search that intelligently interprets the users' search and better understands the business application. Both search implementations were then evaluated by their ability to return the relevant page. The results showed that the new search modeled to the business application did manage to perform better with a higher retrieval success rate and more frequently returned the desired page closer to the first position in the result set.

The purpose of this research study is to examine the effect of public information sources on an individual's satisfaction with the search process undertaken to select a physician. A quasiexperimental research design was adopted to randomly divide the medical staff of a large central Florida medical group into control and intervention groups of approximately 77 physicians each. The intervention involved insertion of the website address to online physician report cards on to each intervention group physician’s profile in the physician directory on the medical group's website. After two months, data were collected consisting of all individuals who had scheduled first-time appointments with one of the medical group's physicians during the two-month intervention period. A random sample of patients was drawn from each group and sample members were mailed a 62-item questionnaire along with a cover letter, summary of the research and postage-paid reply envelope. A total of 706 questionnaires were mailed and 61 completed questionnaires were returned, an 8.64% response rate. Intent-to-treat analysis was conducted using independent-samples t-tests to compare the research study’s continuous variables' mean scores for control and intervention group participants. The analysis revealed no significant difference in scores for control and intervention groups with the exceptions that the control group was somewhat more committed to conducting a search and selecting a new physician. The control group said the physician's communications skills influenced their satisfaction with the search and selection of a new physician quite a lot while the intervention group said physician communication skills somewhat influenced their satisfaction with search and selection. iv Results of the covariance structure analysis demonstrated that information use and level of commitment to search and select a new physician independently predict search satisfaction. As information use and search commitment increase, a patient's satisfaction with the search increases as well. Furthermore, as information use increases, the variety of information sources relied upon or used also increases. The findings support the alternative hypothesis that the positive or direct effect of physician report cards is demonstrated in the time and cost of patient search for a physician for both intervention and control groups. One other alternative hypothesis was partially supported, i.e., the effect of household income is confirmed in patient search and satisfaction in selecting a physician. The alternative hypotheses that proposed that physician report cards are more likely to be used to search for a medical specialist and that physician experience, office location and accepted insurance effect patient search and selection of a physician were not tested. Two other alternative hypotheses were rejected. The research findings also indicated that predictors of health care information search satisfaction vary based upon the environment and contextual factors in which the search is conducted.

No description available.

No description available.

No description available.

No description available.

Arbetet är en empirisk studie gjord för att undersöka beteenden under sökningsprocesser. Dessa processer utgår ifrån den digitala sökningen och mestadels med hjälp av sökmotorer på internet till exempel Google. Olika uppgifter har legat till grund för att kunna sammanställa resultatet inom olika delkategorier. Dessa uppgifter innehåll olika sökuppgifter som användaren ska utföra. Katego-rierna är språket, selektering och mönster. Resultat visar på att användarna använder sig av mo-dellerna Smalna av (Narrow), Omstart (Thrashing) samt Lm1 och Lm2 när de söker. Det är lättast att förklara smalna av modellen med att beskriva den som en filtreringsmodell där personerna utgår ifrån mycket och har kvar det som mest viktigast. Omstarts modell beskrivs bäst som en pro-cess som går tillbaka till sitt startläge men att personerna har andra tankesätt att söka ifrån. De sista två beskriver händelser utifrån agerande. Utifrån de sökresultat användarna fått fram från sina frågor väljer de att analysera de fem första resultaten. Användarna har enbart fokuserat på de län-kar som har liknande innehåll eller namn som deras sökvariabler. Slutresultatet blir att användarna mestadels skiftar sina söktermer eller startar om helt från början efter att de har analyserat vad deras frågor har gett för resultat. När de skriver in sina frågor i sökmotorn använder de sig mesta-dels av enkla nyckelord. Användarnas synsätt på sökmotorerna går inte att fullständigt att fastslå genom denna studie men det är en början. Däremot går det genom resultaten att ge en någorlunda korrekt bild utifrån några personer. / This paper will present results from a study made by a survey with focus of search behavior during search process. These processes is proceeded from the digital information search and mostly with help from search engine like Google. Different tasks have been connected to compile the result into different subcategory. These tasks contains different search tasks for the users to solve. These categories are the language, the selecting and patterns. The result from the survey shows that the users is working with four models and these are Smalna av (Narrow), Omstart (Thrashing), Lm1 and Lm2 when they retrieve information. It is easiest to explain Narrow like a filter model there people start with allot of information and end up with the important information. Thrashing model is best described as an iteration model where people start over but with different search view. The last two models describes events from actions. From the result they get from the information re-trieval they only search from the top five ranked links. The users only focus on the links that have similar content or name like the search variables they used in the search. The end result in this is that the users shifting their search terms or start over from scratch after they have analyzed what result their search question got. When they typing their questions in a search engine they use mostly simple keywords. The users approach how they see a search engine could not be settle trough this thesis but is a start. But from the results it gives an ordinary perspective from some persons.

When searching for an object, we store a mental representation of the target, which guides our search through the use of attention. The effectiveness of this search guidance varies depending on the task and the relationship between target and distractors. With a better understanding of how search guidance changes over time within a trial, we can better compare the differences between experimental conditions. Eye tracking data from a variety of search tasks were analyzed to determine how color guidance varied over the course of the trial. Color guidance for a given fixation was evaluated based on the distance in color space between the nearest object and the target color. These color differences were averaged over all of the trials and plotted based on when the fixation occurred in the trial. The results indicate that color guidance does not begin working at maximum effectiveness immediately. As the trial progresses, the average color difference decreases. After this initial decrease, if the target is not present, guidance becomes less selective and target dissimilar distractors are increasingly fixated. The color distance graphs were compared between experiments to reveal significant differences arising from the experimental conditions.