How Lexxe works
Lexxe Beta version search engine applies the latest Natural Language search technology, which targets two obvious weaknesses of all search engines today.
Lexxe Beta Demo Video (2m46s)
1. Specific information search
Let’s start with an example. If someone wants to find out what colours are associated with Toyota Camry cars, it is common that he or she would type the following three words in a search engine’s query slot:
colour toyota camry
For this type of query, many a time one will feel frustrated, because Google and all other search engines will, at best, search and return documents with a combination of the words “colour” and “toyota camry” in them (both words and/or their variations need to be present in the texts). Most search engines do not know “red” is a color. That’s why the results are all random and inconsistent in quality, meaning the real colour words like “red”, “black”, “blue” may or may not be shown in the snippets. Users cannot fully take advantage of the search results and get the information straight away. The typical frustration from such search experience is due to the missing link between “colour” and “red”, “black, “blue”, etc. By and large in real life, queries of this nature have not been acknowledged and addressed by search engines. Lexxe Search Engine has made a first step towards solving this problem for you with the launch of its Beta version.
We call conceptual words like “colour” a Semantic Key (short for "Semantic Keywords" vs. "Normal Keywords"), which could point to “red”, “black” and “blue”. Semantic keys come in two types: word type and number type based on the format of information it holds. For example, “colour” is a word type with detailed members, like “red”, “green” and “blue” in word form. But the Semantic Key “distance” is a number type, because by nature, it is represented by numbers, plus some words, e.g. 60 Km, 45 miles, etc.
Furthermore, if a reader is familiar with the topic, he or she still needs to read through all the result to have a feel of the result, by spotting those real colour words. When it comes to an unfamiliar topic, spotting the real component (like “red” to “colour”) word will be even harder. With all words in the same appearance, except the key word in bold type, it is hard to pick out information desired.
Lexxe is able to do that in a new fashion. Not only will it return all the results with at least one colour word close to the target search term, but also it will highlight them. Lexxe’s Semantic Keys enable extended search to the elements, which could be basically form unlimited combinations of queries, which will give users a lot of freedom, e.g.
Now, if you think leaving targeted information highlighted in the snippets saves you time without further clicks to open new web pages to locate the information you want, Lexxe has done more for you. Lexxe will also run statistics (upper left corner) on the search result, e.g.
symptom: heart attack
Lexxe’s Semantic Keys actually integrates a kind of knowledge database into search, so professional knowledge like “symptom” could be translated into “pain”, “anxiety”, “depression”, etc. during search. They were highlighted in the search results and their occurrences are computed, because Lexxe knows they are symptoms. This helps a lot, particularly when one is not familiar with the top. It saves heaps of time and energy for users trying to verify the information they receive are correct. No one can have an expert knowledge level of everything anyway. For example, no matter whether one knows the word “anxiety”, it will be recognised as a “Symptom”. Users can basically take for granted that “anxiety” is a symptom. The statistics panel also helps users understand what the most mentioned symptoms are.
Users don’t have to have the knowledge of a lot of the “symptoms” (no one can have an expert knowledge level of everything anyway) to conduct search. The same search will end up in Google with the query words in bold type. While Google’s general results are relevant, its specific information is not pinpointed, making it harder for users to locate and identify information quickly. The stats also help users understand what the most mentioned symptoms are. The amount of time spent could be several times more than using Lexxe depending on how one is familiar with the topic and how fast one reads.
Users don’t have to worry about what Semantic Keys currently available, as the search suggestion or prompt will in most case supply them in a drop-down menu. By typing only a few characters, a Semantic Key might be found in the suggestions, making the use of Lexxe’s Semantic Key feature easier. Lexxe Beta version currently supports about more than 500 Semantic Keys More Semantic Keys are being constructed at Lexxe and it is one of the long term tasks for Lexxe. Users are most welcome to suggest Semantic Keys they want to use in Lexxe search and they will be improved in accuracy in the future.
2. General information search
Opposite to “specific information” search, which is usually made up of a couple of or a few words, is “general information”, which is more than just a few words. “General information” could be anywhere from several sentences or paragraphs to a whole text. General information search intends to return texts (web pages) that are about what the key word means. Google and most other search engines do not analyse text contents in linguistic sense. They took advantage of web page “popularity” information, which are peripheral information naturally existing in web pages on the internet. Whether it is hyperlink reference popularity or hyperlink clicking popularity or traffic information, most search engines play well with them without knowing anything about what is actually and really talked about in the texts.
Lexxe, on the contrary, uses its content knowledge database and linguistic algorithms to identify the real topics of each document. It also judges on the informativity (the degree of being informative) of the texts. It will return documents that are most relevant in content to the query, while popularity only plays a small part, when there is a tie in relevance among web pages. In all environments, intranet or internet or personal data systems, Lexxe tries to return the most relevant documents through its content reading methods powered by its innovative linguistic algorithms and knowledge databases.
Apart from Lexxe Beta’s “specific information search” and “general information search”, Lexxe is also devoted to research and development of a series cutting-edge search technology to provide completely new search experience to the users. Over the next couple of years, Lexxe has planned to debut them for public use step by step.