Türkçe metin tabanlı açık arşivlerde kullanılan dizinleme yönteminin değerlendirilmesi / Evaluation of indexing method used in Turkish text-based open archives

The purpose of this research is to evaluate performance of information retrieval systems designed for open archives, and standards/protocols enabling retrieving and organizing information in open archives. In this regard, an open archive was developed with 2215 text-based documents from "Turkish Librarianship" journal and three different information retrieval systems based on Boolean and Vector Space models were designed in order to evaluate information retrieval performances in the open archive developed. The designed information retrieval systems are: "metadata information retrieval system" (ÜBES) involving indexing with metadata created based only on human, "full-text information retrieval system" (TBES) involving (automatic) indexing based on only machine, and "mixed information retrieval system" (KBES) involving indexing based both on human and machine. Descriptive research method is used to describe the current situation and findings are evaluated based on literature. In order to evaluate performances of information retrieval systems, "precision and recall" and "normalized recall" measurements are made. The following results are found at the end of the research: It is determined that the precision performance of KBES information retrieval system designed for open archives creates statistically significant difference in comparison to ÜBES and TBES. In each information retrieval system, a strong negative correlation is identified between recall and precision, where precision decreases as recall increases. It is determined that the "normalized recall" performance of ÜBES and KBES create statistically significant difference in comparison to TBES. In "normalized recall" performance, no statistically significant difference is identified between ÜBES and KBES. ÜBES is the information retrieval system through which minimum number of relevant and nonrelevant documents; TBES, through which maximum number of nonrelevant and second most relevant documents, and KBES, through which maximum number of relevant and second most nonrelevant documents are retrieved. It is concluded that using OAI-PMH and OAI-ORE protocols together rather than using only OAI-PMH protocol fits the purpose of open archives