关键词: 自发地理信息, 可信度, 质量评价, 信誉度模型

Abstract:

The volunteered geographic information is mostly derived from volunteers' uploading with no quality guarantee. This has become a major obstacle of the application for the volunteered geographic information. Also, the quality becomes the primary problem which requires to be solved firstly. The quality evaluation is the focus of the present research domain. There are many researches on the quality evaluation of volunteered geographic information but a few research on the quality evaluation without reference data. Since it is very difficult or costly to get the reference data, it is important to study the quality evaluation without the reference data. In order to solve this problem with an unknown quality of the volunteered geographic information and the difficulty of getting the reference data with high quality, we proposed the credibility model. The credibility model could evaluate the VGI quality from the number of the volunteers and their reputation for the data changing trend. Then, we turn the qualitative analysis result into the quantitative expression. On one hand, the volunteer's reputation model was built based on this point, meaning a statistic method which takes the proportion for the preserving points as the volunteer's reputation value. Then, the Linus Law adapts and measures the geographical information collected by volunteers, of which credibility relied on the sum of the volunteer's reputation within the research areas. On the other hand, the information quality is gained from the data-changing tendency and measured by the geographical information credibility through computing the degree of change within the research areas. At last, for verification and analyzing the rationality of the credibility model, the OpenStreetMap, collects previous data adapted for Beijing, Shanghai and other cities which require to be used in experiments. Finally, the navigation data is selected as the reference data for the comparison. The result of calculation for the credibility model has a great coherence for the result based on the reference data.

Key words: volunteered geographic information, credibility, quality evaluation, reputation model