Date: Thu, 16 Oct 1997 01:36:51 +0200 From: Wolfram Schneider <wosch@cs.tu-berlin.de> To: www@FreeBSD.ORG Subject: [kim@nexial.nl: Re: Getting NexTrieve] Message-ID: <19971016013651.20542@panke.de>
next in thread | raw e-mail | index | archive | help
-----Forwarded message from Kim Hendrikse <kim@nexial.nl>----- From: Kim Hendrikse <kim@nexial.nl> Message-Id: <199710152231.AAA26340@ns.nexial.nl> Subject: Re: Getting NexTrieve To: wosch@cs.tu-berlin.de (Wolfram Schneider) Date: Thu, 16 Oct 1997 00:31:32 +0200 (MET DST) > Where can I get a copy of NexTrieve for FreeBSD? > > Wolfram We develop it and currently sell it directly. There's two versions of NexTrieve. A lite and a PRO version. The lite version is $1000 USD and the PRO version is $5000 USD. Most people's needs are sufficed with the lite version as the extra speed advantage of the PRO version probably won't be required till after about 50MB of text in a single database. Our pricing is based on a "per machine" price, so it doesn't matter how many databases or virtual domains you index. The PRO version is used by customers like our large newspaper customers (http://www.limbu.nl/cgi-bin/limburger), or our own National web index of New Zealand (www.searchnz.co.nz) etc, i.e. very large indexes. What the lite version is missing from the PRO version is: - Ability to run a text retrieval daemon - Ability to constrain the search with boolean expressions of document properties If you have any further questions, maybe I can best answer them on the phone. -----------------------------------Cut here----------------------------------- NexTrieve - fault-tolerant searching for WWW Introduction NexTrieve is a state-of-the-art "fuzzy" search engine for the web. NexTrieve dramatically improves discovery time by an effective combination of fault-tolerance, improved feedback and page preview functions. Fault-tolerant retrieval NexTrieve's high-speed pattern recognition algorithm is applied to the contents of the "entire phrase". This means that spelling mistakes may be present in any parts of the phrase and NexTrieve will locate the result. A high context sensitivity allows NexTrieve to make effective use of multiple mis-spelt words in context and allow locate so-called "stop words" in context. The importance of feedback We believe that feedback is a critical element in effective text retrieval. Title-only hit lists are not very suitable. Unless the document list is very small, there usually isn't enough information present within the document title to make a good decision. This forces the user to download, possibly useless pages, in order to check visually the relevance of the result. This can be a painful enough situation under normal circumstances, with many providers (At least in Europe) overloading their available bandwidth beyond reasonable limits. One compromise commonly employed is to display the first few lines of the document. This is certainly an improvement, although this only helps if the user query appears in the first few lines of the document. NexTrieve will display "The most similiar couple of lines" to the user query from the document, highlighted of course. In addition, the user has the option of retrieving a "page-preview" of the document. This preview contains much of the markup from the original html document, including all images and functional links. Remember that images also make up valuable part of the context. This facility can save very much valuable time in "qualifying" a hit. The fallacy of stop words One common myth is that certain words, referred to as "stop word", occur too frequently to be useful. The truth of the matter is that words always have a context and this context enhances the meaning of the words in that context. This is why poetry, which is very frequently composed of from "common words", can exist. A very good example is the phrase "To be or not to be" which is comprised entirely of so called stop words. NexTrieve's ranking algorithm can effectively use stop words in their context. Effective enough to bring this classic phrase to the top of the list when searching through the complete words of Shakespeare (http://www.nexial.nl/cgi-bin/shakespeare). Features * Fault-tolerant Incorrect spelling may be present in any part of the query * Highly effective feedback NexTrieve employs an efficient two-stage feedback mechanism, which makes the optimum use of available bandwidth and user patience. * Indexes html, ascii and E-mail With more document formats in the pipeline. * Template-based user interface design Reduces user interface design to a matter of html page layout. Typically a user would build an html page how they want the engine to look and then replace the dynamic components with special tags that are replaced during searching. * Flexible Using using an extensible "tagged" output approach all user-interface and look-and-feel elements are provided outside of NexTrieve. Standard scripts are provided to translate this into a typical user-friendly form, but the user is free to use NexTrieve in many different types of application. * High context sensitivity Which means that even common words (Frequently referred to as stop words) in their context, can be put to effective use. This makes it possible to search for phrases such as the famous "to be or not to be" phrase from Shakespeare. * Small indexes Typically around 20% the size of the original text * Incremental indexing For addition, deletion and update. * Fast indexing NexTrieve will index 100MB of FAQ text in about 13 minutes on a fast PC (Pentium 133Mhz). * Client/Server operation As well as retrieval "on-demand", NexTrieve supports a text-retrieval server for optimum performance on large databases. * Attribute/tags Facilitates searching constrained by the presence of user defined document tags. Named boolean flags and integer tags can be associated with each document and later constrained with an expression during searching. An example may be constraint = date > 19960811 & ( value >= 10000 & value < 20000) & category_a textquery = "laser" In this case the search for the targer "laser" will be constrained to documents that match the above conditions. In addition, the documents alone can be retrieved by constraint alone, providing a "structured" retrieval mode. * Multi-platform Currently NexTrieve is ported to BSDi, SunOS, Solaris, HPUX and SGI IRIX. On-line demos Sounds good? Check it out for your self. We maintain an on-line archive and links to several NexTrieve indexed databases accessable from the page http://www.nexial.nl/search.html. This collection includes several useful databases, mostly relating to Internet technology, along with links to various customer sites. Some of the databases included in this collection are: * Firewalls mailing lists * BSDi-users mailing list * Cisco systems mailing list * Livingston portmaster mailing list * Ascend users mailing list * Hewlett Packard sysadmin mailing list * Usenet FAQ archives * HP Sysadmin mailing list * Sun-managers mailing list * The complete works of Shakespeare Support We pride ourselves in our support and strive to provide the best service to our customers. Most support issues can be handled via E-mail or telephone, but we can provide on-line remote support via telnet if desired and permission is granted. The flexible output format employed by NexTrieve means that it can be used in many and varied environments. Examples include fault tolerant product search as part of an on-line catalogue or electronic shop. If you have some special Internet project, we can provide consulting for solutions or guidelines on to use NexTrieve in your application. Pricing NexTrieve is priced at $5,000 USD/platform. NexTrieve lite is priced at $1000 USD. Support/updates is free for the first three months, after which an Updates/E-mail,telnet,Fax,phone support contract may be purchased for 15%/year. Ordering Send an E-mail to your contact person. They will arrange any details necessary to setup NexTrieve for your particular site. Contact details Physical mail can be addressed to: Nexial Systems St. Annastraat 4 Ohe en Laak 6109 RH The Netherlands Ph : +(31) 475 551643 Fax: +(31) 475 551643 E-mail: sales@nexial.nl -----------------------------------Cut here----------------------------------- - Cheers Kim Hendrikse _____________________________________________________________________________ / \ |Nexial Systems BV. E-mail: kim@nexial.nl | | Ph: +31 475 551643 | | Internet phone: phone.nexial.nl | | (Intel Internet Video phone) | | Fax: +31 475 551552 | |St. Annastraat 4 | |6109 RH | |Ohe en Laak | |The Netherlands | | | |http://www.nexial.nl | \_____________________________________________________________________________/ -----End of forwarded message----- -- Wolfram Schneider <wosch@apfel.de> http://www.apfel.de/~wosch/
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?19971016013651.20542>