Mängden information på internet är enorm och bara forsätter att öka på både gott och ont. Framförallt kan det vara svårt för grupper såsom synskadade och personer med språksvårigheter att navigera sig och ta vara på all denna information. Därmed finns ett behov av väl fungerande sammanfattningsverktyg för dessa, men även för andra människor som snabbt behöver presenteras det viktigaste ur en uppsättning texter. Den här studien undersöker hur väl sammanfattningssystemet CogSum, som är baserat på Random Indexing, presterar med och utan rankningsalgoritmen PageRank aktiverat på nyhetstexter och texter från Försäkringskassan. Utöver detta används sammanfattningssystemet SweSum som en baslinje i undersökningen.

The aim of this thesis is to investigate indexing consistency in LIBRIS, a Swedish national union catalogue, where searches can be carried out among almost five million titles. To calculate consistency, indexing terms of translations and indexing terms of documents in their original language have been compared. I have used a total of 100 such pairs in the study. There is a lot of different methods that can be used when consistency is calculated, in this study Hoopers and Rollings methods have been used since they are two of the most commonly methods used and also have the advantage of being fairly simple to use and understand. In the thesis, differences in consistency between indexing terms consisting of names and general indexing terms and differences in consistency between controlled indexing terms and uncontrolled indexing terms are also compared.

This Master?s Thesis examines the subject indexing of all Swedish gay fiction published from year 2000 until 2007. This is done from a queer theoretical- and domain analytical perspective. The study is conducted in the catalogue BURKsök created by BTJ which is the largest library media company in Sweden. The purpose of the study is to examine the heterosexual norm in fiction indexing and to study to what extent gay fiction is to be found by subject.

The aim of this thesis is to describe the subject indexing of images performed at the provincial museum of Jämtland and to evaluate the inter-indexer consistency at the museum. The main questions are: How are photographs indexed in the database SOFIE foto regarding Outline-codes, subject headings and natural language keywords? How does these procedures compare to research and indexing projects? How consistent are the indexers at the museum when choosing Outline-codes, subject headings and natural language keywords? Answers are sought by reading articles about classification, image indexing, indexing language, indexing consistency and by looking at a number of projects attempting to standardise image indexing practice in Sweden. The indexing practice at the museum is also described. In addition a minor inter-indexing consistency study, based on the subject indexing of ten photographs made by five indexers at the museum, is performed.

This thesis contains an empirical study of exhaustivity in the Swedish National Union Catalogue LIBRIS. The purpose of the study is to examine if fewer numbers of indexing terms in general lead to lower exhaustivity. The study contains two parts, the first part examine the mean of indexing terms per monography in the selection, the second examines the grade of exhaustivity within three groups with a low, moderate or high number of indexing terms. The 55 randomly selected titles were graded on a scale that was created from the indexing theories of Wilfrid F. Lancaster.

The aim of this thesis is to evaluate a Swedish subject headings list for childrens and young adults fiction, from the perspective of consistency, and to analyze this concept's importance as a tool for measuring indexing quality. Another aim is to examine how librarians view the subject headings list as an aid for indexing fiction. The main questions are: How does the degree of consistency change when using a subject headings list for indexing fiction compared to free indexing? How does indexing change according to the facets Form/Genre, Time, Space, Matter and Person when using the list? What do consistency calculations reveal about indexing quality? How do the participants look upon the list as an aid for indexing? The empirical material consists of results from two different indexings of fiction, conducted by librarians. Consistency is calculated using Hoopers and Rollings algorithms.

This thesis analyses the conditions for subject indexing of fiction by studying the relationship between the nature of fiction and the properties of subject indexing. The analysis is founded on a reading of a recently published Swedish subject headings list and uses a theoretical framework of Adornos theory of modern reason. The study is conducted through a qualitative text analysis in three steps. In the first step the subject headings list is analysed by examining how it provides subject access to fiction by the aspects of form, frame and subject. In the second step of the analysis these aspects of subject access are analysed and discussed by relating them to relevant research of subject indexing of fiction.

Randomness is a complex area of computer science, where only pseudo random numbers are possible to generate algorithmically.This report presents established algorithms for random number generation by conducting a basic study of relevant literature, while testing distribution and speed of the algorithms contained within the C++11 Random library. The results are compared to the algorithm XORShift and a modified version of it. The study shows that the modified algorithm passes all tests in Small Crush, andthat the run time is shorter than for those in the Random library. The conclusion is that it is possible to, with simple means, modify an algorithm to outperform the algorithms of the C++11 Random library..

Har ni någon bra kärleksroman? : en studie av indexering av skönlitteratur vid stadsbiblioteken i Malmö och Mölndal. Have you got a good love story? : a study of fiction indexing at the public libraries of Malmö and Mölndal.The thesis analyses interviews with users of two Swedish public libraries to find aspectsof fiction important to users and consequently to be considered when indexing fiction.Descriptions of a novel by 50 users at each library are compared to the index terms usedfor these novels. For the analysis, the authors propose a model of their own based onAnnelise Mark Pejtesen's AMP system for fiction indexing.

The aim for this master?s thesis is to investigate the indexing consistency in a folksonomy Delicious. This is done by investigating how users construct tags. By applying a semiotic perspective and use analysis of content the occurrences of inflected forms, hyponymy, meronymy and users own tag creations were established. Lancaster?s seven factors is the starting point for the discussion about indexing and the factors that affects it in either a negative or positive way.

The aim of this thesis is to investigate how the indexing of fiction for adults looks like in the public libraries that don?t buy indexed records for fiction from Bibliotekstjänst (BTJ). The questions that I want these libraries to answer is: - Public libraries that index fiction for adults themselves: 1. What tools do they use when indexing?, 2.

This is an explorative study about subject indexing of a special collection of books, administered by the John Andrén Foundation at the library of Ystad. The study summarises a number of theoretical perspectives as well as some practical instructions on how to carry out the task. Theories have been placed into five different groups according to their various approaches to subject indexing. These have been analysed in regard to their usefulness to the Andrén collection. The authors have tested and evaluated six practical models on a selection of books from the Andrén collection.

The web has made an incredible amount of unorganized information available to anyone. There are search engines that help us structuring the information, but it is still difficult to find what you search for on the web. The purpose of this master's thesis is to investigate whether the already existing techniques for automatic indexing are suited for the new information retrieval context on the web and how the choice to support these techniques, or not, affects the search results. This is examined through a literature study on automatic indexing and other related concepts, such as information retrieval and information searching on the web in order to get a theoretical frame to the work and by an observation of the search engine HotBot to approach the purpose of the thesis. The observation is carried out by searching HotBot's database and investigating the search results in order to try to identify patterns that can reveal something about how HotBot's automatic indexing is done.

Social indexing is an approach to manual indexing of web content. The purpose of this master thesis has been to investigate terminological and conceptual consistency in social indexing of web content. Conceptual consistency was investigated/measured focusing on synonyms and (the use of) singular and plural forms of words. In order to measure consistency in social indexing data were obtained from the Delicious website. The data set consists of five indexed objects, 3001 tags and 239 users.

The aim of this thesis is to examine possibilities concerning subject access to fiction film, i.e. appropriate ways to describe fiction film for subject retrieval. Problems regarding subject indexing in general, and subject indexing of fiction film in particular, are identified, as well as the uniqueness of fiction film and film genres. Four types of subject aspects are discussed: form and genre, fictitious characters, setting, and general topics. Several guidelines, form- and genre lists, and subject headings lists are examined and treated as potential indexing tools.