Computerized system and method for high-quality and high-ranking digital content discovery

A non-transitory computer-readable storage medium tangibly encoded with computer-executable instructions, that when executed by a processor associated with a computing device, performs a method comprising:

• receiving a request for image content;
• identifying a set of candidate images, said identifying comprising searching a database of images and identifying images that correspond to the requested image content;
• translating, using vector analysis software, each identified candidate image into a feature vector, said feature vector comprising information associated with features of each candidate image;
• applying a ranking function to each feature vector, said application comprising calculating a score for each candidate image based on the ranking function being applied to the feature vector of each candidate image, said ranking function comprising a combined logistic loss function and pair-wise loss function, said application of the ranking function further comprising simultaneously determining a quality value and relevance value for each candidate image via the combined logistic loss function and pair-wise function of the ranking function;
• determining query-image pairs based on the received request and the identified candidate images, each query-image pair comprising information associated with the received request and an individual candidate image from the set of candidate images;
• assigning said calculated score for a candidate image to its respective query-image pair;
• ranking the set of candidate images based on the scores of the applied ranking function, wherein candidate images with higher scores are ranked higher in said set than those with lower scores;
• and
• communicating, over the network, a search result to a user in response to said request for image content, said search result comprising said ranked set of candidate images.

A system comprising:

• a processor;
• and
• a non-transitory computer-readable storage medium for tangibly storing thereon program logic for execution by the processor, the program logic comprising:
• logic executed by the processor for receiving a request for image content;
• logic executed by the processor for identifying a set of candidate images, said identifying comprising searching a database of images and identifying images that correspond to the requested image content;
• logic executed by the processor for translating, using vector analysis software, each identified candidate image into a feature vector, said feature vector comprising information associated with features of each candidate image;
• logic executed by the processor for applying a ranking function to each feature vector, said application comprising calculating a score for each candidate image based on the ranking function being applied to the feature vector of each candidate image, said ranking function comprising a combined logistic loss function and pair-wise loss function, said application of the ranking function further comprising simultaneously determining a quality value and relevance value for each candidate image via the combined logistic loss function and pair-wise function of the ranking function;
• logic executed by the processor for determining query-image pairs based on the received request and the identified candidate images, each query-image pair comprising information associated with the received request and an individual candidate image from the set of candidate images;
• logic executed by the processor for assigning said calculated score for a candidate image to its respective query-image pair;
• logic executed by the processor for ranking the set of candidate images based on the scores of the applied ranking function, wherein candidate images with higher scores are ranked higher in said set than those with lower scores;
• and
• logic executed by the processor for communicating, over the network, a search result to a user in response to said request for image content, said search result comprising said ranked set of candidate images.