A Proposed Method for Semantic Annotation on Social Media Images

Semantic annotation attaches to various concepts (e.g. people, things, places, organizations etc) in any other content. A semantically annotated document is easy to interpret, combine and reuse by computers. In human history Web plays as the greatest information source. Many researchers believe semantic annotations can be inserted in web-based documents for information extraction and knowledge mining. These annotations use terms defined in an ontology. This paper uses the annotation process based on textual patterns for improving annotation. Introduction Image databases are large and databases are used in public domain applications and domain-specific applications. Internet services such as social networks, photo sharing image and video monitoring are the specific examples. Specific-domain applications use medicine, aerial surveillance, audiovisual archiving and many more. Many image semantic annotation approaches are the automatic association between low-level or mid-level visual features using machine learning techniques. Machine learning is insufficient to bridge the well-known semantic gap problem for achieving efficient systems for automatic image annotation. Structured knowledge models, such as semantic hierarchies and ontologies, will be a good way to improve such approaches. These semantic structures allow modeling many valuable semantic relations between concepts such as contextual and spatial relationships. These relationships are of prime importance for the understanding of image semantics. Structured knowledge models about high-level concepts can reduce the complexity of the large-scale image annotation problem. The basic approach is prepare a methodology for building and using structured knowledge models for automatic image annotation. Initially it deals with the automatic building of explicit and structured knowledge models, such as semantic hierarchies and multimedia ontologies, dedicated to image annotation. It helps for building semantic hierarchies faithful to image semantics. Imagesemantic utilizes to find out the similarity measure between concepts and on a set of rules for building of the final hierarchy. It goes further in the modeling of image semantics through the building of explicit knowledge models to incorporate semantic relationships between image concepts. The proposed method creates automatically building multimedia ontologies consisting of subsumption relationships between image concepts, and also other semantic relationships such as contextual and spatial relations. The problem of finding a desired image (or a subset of them) becomes critical as the digital images increases in social media. In a distributed medical database. Given the image of a patient, it can be find other images of the same modality, of the same anatomic region, and/or of the same disease that was already diagnosed and it can help on the clinical decision-making process. Multimedia image retrieval is still a big challenge. The image retrieval field is carrying on the set of techniques/systems for browsing, searching and retrieving images from a large collection of digital images. Systems operate in two phases: i) image indexing: which could be defined as the process of extracting, modeling and storing the content of the image, the image data relationships, DOI: 10.18535/ijecs/v6i6.31 Mrs. Sayantani Ghosh, IJECS Volume 6 Issue 6 June, 2017 Page No. 21737-21742 Page 21738 or other patterns not explicitly stored, and ii) image search: which consists in executing a matching model to evaluate the relevance of previously indexed images with the user query. Figure 1 illustrates the image retrieval systems. It is common sense that image search in these system is based on the same features (or modalities) than the ones used for image indexing. Figure 1 Workflow of image retrieval systems. Automatic image annotation is a process of automatically assigning a text description (reduced to a set of semantic keywords) to a digital image through a computational model. Automatic image annotation means image retrieval systems in order to index and retrieve images of interest from a large database. This task is regarded as an image classification problem described by the following steps: 1. Training image dataset consisting of a set of images with their textual annotations is collected. These textual annotations consist of semantic concepts depicting image content. The set of all concepts composed the annotation vocabulary. 2. Computational model enabling to find a correspondence model between the low-level or mid-level representations of images is framed for the annotation vocabulary. 3. Test the system and adjusting the parameters of the computational model. This paper deals with the problem of semantic image annotation. It focuses on how to model in an effective way to find the image content. The approach is based on the building and the use of explicit and structured knowledge models in order to improve image annotation.