Group items tagged

"The Web puzzle of online information resources often hinders end-users from effective and efficient access to these resources. Clustering resources into appropriate subject-based groupings may help alleviate these difficulties, but will it work with heterogeneous material? The University of Michigan and the University of California Irvine joined forces to test automatically enhancing metadata records using the Topic Modeling algorithm on the varied OAIster corpus. We created labels for the resulting clusters of metadata records, matched the clusters to an in-house classification system, and developed a prototype that would showcase methods for search and retrieval using the enhanced records. Results indicated that while the algorithm was somewhat time-intensive to run and using a local classification scheme had its drawbacks, precise clustering of records was achieved and the prototype interface proved that faceted classification could be powerful in helping end-users find resources. "

"Summon Search API uses an authentication scheme based on secret key verification via an HMAC-SHA1 digest. The API performs both authentication and authorization via the same HTTP Authorization header. Creating an Authorization header for an API request requires an access ID, a paired secret key, a client key, and access to certain elements of the HTTP request being authenticated. To create the Authorization header, specific request elements must be assembled into a string that uniquely identifies the request. This ID string is then turned into a digest using the HMAC-SHA1 algorithm as defined in RFC 2104. The digest is then Base64-encoded according to RFC 2045, and assembled into a header along with the access ID. When the API server receives a request, the first thing it does is check the x-summon-date header to make sure it is within a reasonable margin of the server time. If the header time is not within one hour of the server time then authentication fails immediately. If the request timestamp is acceptable, then the server goes on to check that Summon authentication is being used in the Authorization header. If Summon authentication is being used, the server extracts the access ID, client key and encoded authentication digest from the Authorization header, and looks up the paired secret key using the access ID. Finally the server performs the same algorithm described in the previous paragraph to produce a test digest, and compares the test digest against the digest provided in the Authorization header. If the two digests are identical, the request is authenticated."

"(...) the relevancy ranking within Summon is based on term frequencies as Aaron said, but many other algorithms are used as well, such as field weightings and inverse term frequencies to counterbalance any skewing from the full text content. Additionally, each record in Summon has a static rank that allows for balancing records from one collection against another to ensure all records are treated equally. "

"Web Search engines most often operate on the idea that their tools are supposed to represent the fastest way between a user's search query and a site that will represent a fitting information content to satisfy a "information need". The main challenge is therefore about building ranking algorithms that will find the "best" site and put it into first place. While this approach has been highly successful it also neglects the user's ability to interact, select, navigate, analyze - in short, to be an active agent and a stakeholder in the search process. Our proposal is the technical manifestation of a simple concept: instead of reducing the search phase to a minimum we believe that there is much to be gained by presenting a semantically rich interface to users; this interface can be used to either navigate in a list of results (1000 in the case of our current prototype) or to refine the search query. The "termcloud" and "cluster" views give a quick overview of a topic by showing which terms are significant, their relative frequency and the relations that exist between them. This way the user gets an idea of the topic before even visiting a Web site from the result list. We hope that this approach will help users in digging deeper than the first couple of results and explore the regions of the Web that might not be so easy to find with a regular, hit-driven search engine."

"This is going to be the first of a couple of posts related to Voronoi Tessellations, Centroidal Voronoi Tessellations and Voronoi TreeMaps. In this post I'll explain what a Voronoi Tessellation is, what can it be used for, and also I'll describe an interesting algorithm for creating a Voronoi Tessellation given a set of points (or sites as I'll call them from now on)."

" VuFind is a library resource portal designed and developed for libraries by libraries. The goal of VuFind is to enable your users to search and browse through all of your library's resources by replacing the traditional OPAC to include: * Catalog Records * Locally Cached Journals * Digital Library Items * Institutional Repository * Institutional Bibliography * Other Library Collections and Resources VuFind is completely modular so you can implement just the basic system, or all of the components. And since it's open source, you can modify the modules to best fit your need or you can add new modules to extend your resource offerings. VuFind has many APIs to interact with the search, data and many other features. You can syndicate your record data with other institutions via an OAI server. You can search using vufind's algorithms via OpenSearch. And if you want complete access to your indexed data, you can interact with Solr, VuFind's backend search and index engine."