Our Researches | Solutionsbeyond


		Our Researches Solutions Beyond believes in stunning people with the quality and kind of Researches which we do. Super solutions to challenging demands is our biggest strength and the excellence that we achieve in doing path breaking researches is the company's core value.
	Heart of a Search Engine Category : Grid Computing, Artificial Intelligence > Natural Language Processing Context Detection Engine (Codename) is like the heart of a small scale search engine. It's an algorithm that looks at the webpages exactly the way search engines do and translates the webpages into simple keywords and clusters to make them easy to categorize and crawl. The results of this algorithm match with the results of some major search engines (upto 90% accuracy) and easily predicts the keywords on which a landing page might be ranked. Same will be the heart of a small scale search engine that Solutions Beyond is working on.
	Product Specification grabber Category : Artificial Intelligence > Natural Language Processing Algorithm had to scan English literature written on webpages and dig characteristics of a product that it was trained to understand. For example, you could train the algorithm to find characteristics of "Hosting Space" sold by thousands of web hosting companies around the web. The algorithm will visit webpages of hundreds and thousands of web hosting companies and find characteristics such as "Monthly Price, Disk Space, Bandwidth, Setup Price, Control Panel used" and more from the products page of Web Hosting companies. Likewise you can train this algorithm to intelligently find characteristics of thousands of "Circuit Diagrams" offered on hundreds of electronics -websites, and search for circuit diagrams that "solve a specific purpose, are sold between a specific price range, and run on a specific range of input voltage".
	Internet Proxy with Intellectual Property Protection Category : Network Security Project demanded that office employees routed from a proxy should be able to visit and access any third party/untrusted website, but should Not be able to upload confidential data while browsing any untrusted website or server. An Internet Proxy software was modified to support this "Intellectual Property Protection" component. With this component attached, the Proxy Server began to understand signature of file-uploads inside packets and was able to : Break "Attach-File feature" of all webmail websites(like gmail, yahoomail, any even custom webmail scripts that are created from scratch) Break "Upload File feature" of all websites which allowed visitors to choose file from their Hard disk and upload on website's webserver (like yahoo briefcase, rapid share, even the custom file-upload scripts created from scratch) Not just the packets of Internet Browsers, but the The Intellectual Property Protection (IPP) component was architected to understand and moderate packets of any "Internet Enabled Application" that runs on Proxy-user's computer (for example Instant Messengers, intranet FTP Servers etc). IPP component allows all Internet Enabled Applications to operate without any interruption, and used to block their packets only when IPP sniffed some file-transfer or suspiciously confidential data being transmitted by Application. IPP used to judge bad-packets of Application from good-packets of Application without having trained on the protocol that Application uses.
Keyword Categorizer Category : Artificial Intelligence > Natural Language Processing Internet Marketers need to deal with 1000s of keywords everyday to make their website gain more train or earn more income. An uncategorized list of random 1000 keywords makes Internet Marketer's decision making process very slow. This algorithm when given a list of 1000s of keywords, tries to deduce meaning of each keyword and pick parent-keywords from the same keyword list and categorize remaining keywords in the keyword list as child of the parent keywords. This way the entire keyword list of 1000s of keywords could be summarized in just 10-20 parent keywords (with rest keywords shown under the parent keywords when parent keyword was double-clicked). Once the list of 1000s of random keywords is categorized in categories, the Internet Marketers are able to compare entire Group of keywords with a different Group of keywords and save time that they would have spent by comparing each keyword one by one with keywords in entire list Visit the Keyword Country Labs for a live demonstration
Synonymous, but for 2-worded or 3-worded keywords Category : Artificial Intelligence > Natural Language Processing Finding synonyms of single-words are easy. But here our algorithm was required to 'derive' synonyms of 2-word or 3-word keywords that were not otherwise found present Thesaurus. Phonetic Misspellings Calculator Category : Artificial Intelligence > Natural Language Processing Given a keyword this algorithm will analyze how the words sound Phonetically and suggest misspellings that people might type while searching for that keyword in search engines. This misspelling calculation algorithm was further used in Search engine scripts to help Search Engine determine whether user typed something wrong. The list of misspellings was also used by Internet Marketers to survey the popularity of misspelled keywords. Unlike the traditional method of Keyword Misspelling Calculation (which used to check location of each character on keyboard and generate list of keyboard typing mistakes that user might type), this new algorithm takes a new approach by analyzing phonetics of each character used in words to generate list of misspellings for each ethnicity. Because words are pronounced differently in different ethnicities, hence this algorithm could predict whether Asians have more probability of searching with a certain misspelling, or Latin Americans, or British. Examples of misspellings that this algorithm generates when you input "Britney Spears" in it : Showing 1-10 of 357 misspellings britaney spears brytney spears britny spears bretney spears brittny spears britnay spears briteney spears brittiny spears brittny spears britiney spears Anti Password Sniffing on HTTPS Category : Network Security The project involved automatically filling Username & Password on secure (httpS) webpages that User has opened in his Internet Explorer. It was important for our application (installed on user's Internet Explorer) to fill the password without letting the user know what password was filled A fundamental problem with preventing the User from knowing password that was filled is that User could easily figure out passwords filled in his Internet Explorer by using any of the "Password-Textbox revealing tool" or "Network Packet Sniffing tool" available on internet (one such tool can be downloaded from here). To solve this fundamental problem, this "Anti Password Sniffing" algorithm was devised, which returns a hoax password to Password-Revealing Software or Packet Sniffing Software that tried to steal passwords filled by our password-filler application ..to count a few Above are only the researches that we have been working on after moving into product development. There are many startling researches done during the time when we were catering to outsourcing sector, about three years back. However, these cannot be revealed as we believe in honouring Non-Disclosure Agreements.