bachelor's thesis
Aleksandar Cuculoski (Author), Veljko Pejović (Mentor)

Abstract

Mobile phones are a popular platform that has a close relationship with the user. Such a device acts as an intermediate between the user and the applications that keep us socially active and help us schedule our tasks, shop, or navigate in our everyday lives. As such, these applications require a portion of our time, which, on the other hand, gives them the ability to request user's attention through notifications. However, this ability does not necessarily mean reachability as that notification might arrive at an inappropriate time, thus having a negative impact on the user, resulting in poor task performance, stress, and annoyance. In the following thesis, we sought to determine the most appropriate situations of interrupting users without creating a heavy burden on the phone. Through a real-world mobile-app data collection campaign with 19 volunteers over 2 weeks we collected fine-grain sensor information about a user’s context. From the features extracted we created machine learning models to determine the most informative sensors. We used that data to chart sensors' informativeness versus their energy consumption. We then devised a method that allows a controlled trade-off between the accuracy of the interruptibility inference and energy consumption, as such allowing for a specific energy-optimal interruptibility management.

Keywords

mobile sensing;interruptibiliy;approximate mobile computing;computer and information science;diploma thesis;

Data

Language: English
Year of publishing:
Typology: 2.11 - Undergraduate Thesis
Organization: UL FRI - Faculty of Computer and Information Science
Publisher: [A. Cuculoski]
UDC: 004.7:004.5(043.2)
COBISS: 33442819 Link will open in a new window
Views: 768
Downloads: 140
Average score: 0 (0 votes)
Metadata: JSON JSON-RDF JSON-LD TURTLE N-TRIPLES XML RDFA MICRODATA DC-XML DC-RDF RDF

Other data

Secondary language: Slovenian
Secondary title: Energijsko učinkovito inteligentno upravljanje mobilnih obvestil
Secondary abstract: Mobilni telefoni so vsepovsod prisotne naprave, ki so z uporabnikom tesno povezane. Delujejo kot vmesnik med uporabnikom in aplikacijami, ki uporabnika ohranjajo družbeno aktivnega in mu pomagajo načrtovati opravila, nakupovati ali krmariti v vsakdanjem življenju. Aplikacije po eni strani zahtevajo čas in pozornost uporabnika, po drugi strani pa pripomorejo k njegovi dosegljivosti. Vendar to ni nujno pozitivno, saj lahko obvestilo prispe v neprimernem času, kar negativno vpliva na uporabnika. To lahko povzroči slabše izvajanje nalog, stres in druge nevšečnosti. V diplomskem delu smo skušali ugotoviti katere so najustreznejše situacije za prekinitev uporabnikov, ne da bi pri tem porabil veliko baterije. S kampanjo zbiranja podatkov iz mobilnih aplikacij devetnajstih prostovoljcev smo dva tedna zbirali podatke, ki se nanašajo na uporabo aplikacij v danem kontekstu. Ustvarili smo modele strojnega učenja in določili najbolj informativne senzorje za namen ugotovljananja prekinljivosti uporabnika. S pomočjo podatkov smo načrtovali informativnosti senzorjev glede na njihovo porabo energije. Nato smo zasnovali metodo, ki omogoča nadzorovan kompromis med točnostjo sklepanja o prekinljivosti in porabo energije, oz. omogoča energijsko optimalno upravljanje prekinljivosti.
Secondary keywords: mobilno zaznavanje;prekinljivost;približno mobilno računalništvo;računalništvo in informatika;univerzitetni študij;diplomske naloge;
Type (COBISS): Bachelor thesis/paper
Study programme: 1000468
Embargo end date (OpenAIRE): 1970-01-01
Thesis comment: Univ. v Ljubljani, Fak. za računalništvo in informatiko
Pages: 51 str.
ID: 11954091