كليدواژه :
نقشه كروكي , روابط مكاني كيفي , خوشه بندي اماكن , انطباق معابر
چكيده فارسي :
يكي از شناخته شده ترين روشهاي مورد استفاده توسط عموم مردم براي به اشتراك گذاري دانش مكاني، كروكيها هستند. كروكيها، با به تصوير كشيدن بخشي از شناخت مكاني افراد از محيط پيرامون ميتوانند به عنوان روشي كارا در زمينه جمع آوري داده هاي مكاني مورد توجه قرار گيرند. با توجه به اينكه هريك از كروكيهاي ترسيم شده از يك منطقه مشخص در برگيرنده اطلاعاتي متفاوت است از اين رو ادغام اين كروكيها ميتواند منجر به تهيه كروكي كاملتري در مقايسه با هريك از اين كروكيها شود. مساله مهم در اينجا يافتن تطابق بين عوارض موجود در كروكيها به عنوان مبناي فرآيند ادغام ميباشد. بر اين اساس در اين مقاله راهكار جديدي براي انطباق بين معابر موجود در كروكيها ارائه ميگردد. از آنجا كه مفهوم مجاورت بين عوارض در شناخت مكاني افراد از اهميت بالايي برخوردار است؛ از اين رو ارائه اين راهكار، با تمركز بر مفهوم مجاورت و نزديكي و نيز اطلاعات توصيفي اماكن و معابر موجود در كروكيها صورت گرفته است. در پيادهسازي راهكار مورد نظر، فرآيند انطباق چهار كروكي به صورت دو به دو انجام شده است. به منظور ارزيابي دقت اين روش مقادير Precision و Recall براي هر انطباق بين دو كروكي محاسبه شده است. مقادير متوسط اين دو پارامتر (به ترتيب 89.63% و 87.1%) نشان ميدهد كه اين راهكار در مقايسه با ديگر پژوهش‌ها انجام شده در اين حوزه، از دقت قابل قبولي برخوردار بوده است. از نتايج اين طرح ميتوان در زمينه تكميل اطلاعات توصيفي و هندسي در نقشه هاي متريك با انطباق با كروكيها و نيز در سيستمهاي پرس و پاسخ از پايگاه داده مكاني با استفاده از ترسيم توسط كاربران استفاده كرد.
چكيده لاتين :
One of the most well-known methods used by the public to share spatial knowledge is the sketch maps. These types of maps can be considered as an effective way to collect and share spatial data by depicting some of the spatial cognition of individuals from the real world. Because of the difference in the cognition of individuals from the environment, each of the sketch maps drawn from a specific region can include different information. An important question to be asked here is how the potential of these maps can be benefited more and better in order to collect spatial data in spite of these differences. One of the proposed methods for this purpose is to combine and integrate information extracted from different sketch maps of a region with the aim of obtaining a map with more complete information than any of the original sketches. Although this integration can provide a better use of the potential of the sketch maps for collecting spatial data, prior to that, it is necessary to consider the matching process between the features in the sketch maps. However, there are several types of features in sketch maps, but all sketches contain pathways and these pathways can be identified by name, or based on specific places or Points of Interest (POI) located in the proximity of them. Accordingly, in this paper, a new solution is proposed for matching between the existing pathways in the sketch maps. In order to describe each pathway in the sketch maps, in addition to the name of the pathway, we can use information that can be extracted indirectly from these sketch maps. To describe a specific address, even if we do not know the correct names of the pathways, we can use the names of near POIs or other related pathways. The same point is used in this paper to describe the pathways in the sketch Maps. Accordingly, the matching parameters are considered as two main parts. These parts are the proximity of POIs and intersections of pathways with matched well-known pathways. Therefore, in the first step of implementation, based on the descriptive information as well as the relative distance between the POIs, the clustering process is performed on POIs location. Subsequently, based on the qualitative spatial relations between clusters and pathways, descriptive information of the pathways, the arrangement of the pathways around the intersection points, pathway intersection with roundabouts and matched well-known pathways, the process of matching of the four sketch maps is done. In order to evaluate the accuracy of this method, the outputs of the matching process are compared with the manually matching, and then, the precision and recall values for each matching are calculated. The average values of these two parameters (89.63% for precision and 87.1% for recall) indicate that this solution has an acceptable accuracy compared to other studies in this field and therefore, the proposed solution can be exploited in future works as the basis for the integration to enrich the existing metric maps with less details.
