Quepasa Announces Agreement to Acquire Vayala Corp.; Represents Quepasa's Return to Online Search Services for the U.S. Hispanic Market and Entry to the Latin Search Marketplace.

Business Editors

PHOENIX--(BUSINESS WIRE)--Aug. 22, 2002

Quepasa.com Inc. today announced a definitive agreement to acquire privately held Vayala Corp. of Phoenix.

Vayala is a developer of intelligent search and indexing technologies for the public Internet.

This acquisition represents Quepasa's return to Internet search services for the U.S. Hispanic market, the fastest growing ethnic group online, according to market research firm Nielsen/Netratings in a study released July 16, 2002. In addition, this acquisition represents Quepasa's entry to the international Spanish language search marketplace.

In connection with the agreement, Quepasa will issue approximately 32 million shares of its common stock, and options to purchase an additional 65 million shares of common stock, to Vayala's shareholders. As part of the agreement, Quepasa's board of directors have agreed to effect a reverse split of one share for each 20 shares of quepasa stock outstanding.

Based on a recent closing price of .050 cents per share of Quepasa common stock, this transaction currently represents market value of approximately $5 million.

With Vayala's technology, Quepasa as a business will, for the first time, be able to operate its own proprietary database of Internet search results. In the past, Quepasa had licensed its search results from Inktomi Corp. (Nasdaq: INKT).

In addition, Quepasa will be able to extend its business model and offer its search results for licensing to others, to meet the evolving needs of the rapidly growing online Latino population.

In addition, Vayala's technology introduces new intelligent services that significantly reduce the costs associated with, and establish a benchmark for "private-label" Internet search products which have not historically been available to both the domestic and international Spanish language online community.

About Vayala Corp.

Vayala Corp. ("Vayala"), incorporated in July 2001, is an Arizona company that specializes in developing and marketing dynamic search technologies to portals, wired businesses, and other clientele who require access to dynamic, relevant information.

Dynamic search technologies include those solutions that are capable of archiving and maintaining large, searchable databases that are updated in real time or near real time.

Vayala was founded in 2001 by Albert Chen, a Southern California entrepreneur and former founder of SonicScream Corp., and Brian Lu, entrepreneur and eldest son of Asian technology entrepreneur Hong Liang Lu, current president of UT Starcom Inc. (Nasdaq: UTSI).

Principal investors and shareholders of Vayala include members of the Marriott family from Salt Lake City, affiliates of Softbank Corp. (China), Jeffrey Peterson, original founder and CEO of Quepasa.com Inc., Vayala's founders, Albert Chen and Brian Lu, and award-winning motion picture producer Paul Mazursky, among others.

Vayala Search Technology

Vayala Corp. has developed a highly efficient systems architecture for crawling, indexing, and searching that uses a hybrid approach to maintaining a current document index. Vayala's search database is currently capable of the rapid indexing and retrieval of up to 500 million digital documents, and is rapidly scalable to host up to 5 billion documents.

In short, the system maintains a large static index, and a smaller, dynamic, "B-tree"-based index. This section sets forth a brief summary of the Vayala search technology, which currently consists of several independent modules.

The crawling agent for the system is entirely proprietary and based around a network client model. The crawler respects robots.txt files, is capable of thousands of simultaneous connections, and language filtering via dipthong probability analysis.

Vayala technology is also responsible for storing, compressing, and indexing( by unique id and URL hash )digital documents. Also entirely proprietary, it is written and based upon a threaded network server model. The module uses zlib for compression and several flat ISAM files for indexing by document. The module also includes an internal lexicon for resolving words to unique ids.

Vayala technology also employs a distributed application that serves to create inverted indexes of document-to-word mappings, also proprietary, and also based upon a threaded network server design. Using a hybrid static/dynamic model, this technology separates its indices into long-term ISAM-indexed word-to-document mapping runs and short-term b-tree-indexed runs for most frequently followed URLs.

This software uses feedback from end-users or other distinctive seeding data to intelligently and effectively separate documents between the two index types.

Finally, Vayala technology aggregates the results of incoming queries, caches frequent query results, and returns results to querying clients, by XML stream; utilizing a six-pass re-ranking sieve that progressively filters out redundant or irrelevant documents, to adjust rankings based on frequency and proximity analysis.

The Vayala "Seek" technology has capability to influence the indexing of future documents( via RPC mechanism )based on feedback trends.

Technical Data

All communication between processes and machines occurs via XML-RPC or disk-backed FIFO queues. The decoupling effect of Vayala's internal queues allows the system as a whole to operate more resiliently to, for example, a temporary outage of a single host.

The system uses about 20 different ranking methods, ranging from word importance analysis, to proximity analysis and categorizing algorithms. Rank is a value that is adjusted at every step of the process. All modules have a significant influence in the ranking of results. The job is made significantly easier if the document domain is restricted to a specific language or region.

The system is designed to capitalize primarily on its ability to focus on a particular language domain, which makes relevancy easier and more effectively managed.

The logical components comprising Vayala Search are all designed to run on dedicated hardware, with the exception of Vayala's "Fury" component. Fury is a distributed application, with each instance responsible for a specific range of document and word id. Vayala Seek may sometimes also run on the Fury hardware, alongside Fury.

The entire service was designed and built to run on commodity hardware running the Linux operating system. The system is horizontally scalable to support larger query volume by mirroring Fury and other machines. The system is vertically scalable to support larger indexes OR more frequent crawls by increasing internal bandwidth and processing power on the dedicated machines.

The Vayala product is network-intensive, and therefore the network server models are extremely important. For most purposes, the threading capabilities of current Linux kernels are fast and solid, with minimal context switching compared with earlier kernels. Vayala therefore uses threaded network servers throughout the system.

The XFS filesystem from SGI provides exactly the kind of high-performance extent-based services that our technology calls for, particularly for writing the ISAM indexes of Fury. The open-source XML parser Xerces is integral to RPC mechanisms across servers and for formatting search results for returning to the end-user.

The system's reliance on mature, tested, open-source technologies gives a compelling value proposition in that Vayala can make custom changes to the system on demand, even at the operating system level, if needed.

All resource access is controlled by abstraction layers. C++ class libraries have been written for all access to disk and network layers to allow for possible future portability and to help bridge 64-bit filesystem compatibility issues. All technologies has been custom developed by Vayala.

Vayala hosts use internal RAID arrays for primary repository storage, which then store all redundant data. The lexicon is a bit-for-bit image of the heap space used by the lexicon module, flushed to disk by a reentrant function run periodically by a dedicated heartbeat thread. Vayala technology stores compressed document text, and indexes of each document as coded data.

Unique and innovative to the search market, Vayala's "Fury" technology maintains two concurrent static indexes: one that is marked "active", and another that is marked "developing". The active index is specified to exist on a disk or array that is physically distinct from that of the developing index. As document data arrives, the developing index is written.

On predefined intervals, the developing index is merged with the active index, ranked, sorted, and written back to disk. The former developing index then becomes the current active index, and the machine state returns to its former state, with the active and development roles reversed. These files are represented in pairs.

Indexes

The first priority of the Vayala search product is to keep disk seek counts to a minimum across the entire system, and cache information wherever possible. The indexing strategy is key to achieving this. Multiple indexes are maintained across the entire system: Attic ISAM, Document Hash ISAM, Active Fury ISAM, Developing Fury ISAM, and Dynamic Fury Index.

The Attic ISAM is a normal ISAM file that indexes into the main document text repository. This index ensures single-seek access to document data in the worst case. As the system caches large portions of this ISAM in memory, often a single seek is not even required.

The Document Hash ISAM is an additional ISAM that serves to resolve document URL hashes to document IDs and indexes into the main text repository. This allows administrators to resolve specific URLs for deletion or alteration within the system, without requiring sequential scans through the entire repository data.

Unlike the Attic ISAM, the Document Hash ISAM is implemented as a sparse file, with an addressable space of 2^128, since URL hashes are 128 bits.

The Active Fury ISAM and Developing Fury ISAM are structured identically, with the exception that the Active index is ranked and sorted, while the developing index has yet to do so. The Dynamic Fury Index is a B+-Tree structure that stores word-to-document mapping data for those few documents which are refreshed in near-real-time.

The next revision will see this structure replaced with a large, sorted array over which a binary search will operate, with the caveat that the hosting hardware must supply enough RAM for the operation.

Performance

Vayala's proprietary technology is highly efficient. Each of the services described use buffering, caching, and load-balancing of request volume wherever possible. However, despite the amount of data cached and buffered in ram, the size of the entire data set relative to the amount cacheable data is very large.

The system uses adaptive cache algorithms that help to decrease the cache miss rate, all of which ultimately require statistical feedback, such as most frequently-accessed documents, most frequently-requested search terms, etc.

The need to balance probabilistic caches with the ever-changing nature of the "shallow-web" index managed by the Dynamic Fury Index noted above makes the problem minimal.

Vayala's focused Spanish-language search product will become available for end-user usage in Q4 2002, with beta testing beginning in late Q3 2002.

About Quepasa Corp.

Quepasa.com Inc. is a bilingual (Spanish/English) Internet portal and online community focused on the United States Hispanic market. Quepasa provides users with information and content centered around the Spanish language. Because the language preference of many U.S. Hispanics is English, we also offer our users the ability to access information in the English language.

Quepasa.com, Quepasa, and the Quepasa.com logo are registered trademarks of Quepasa.com Inc. and/or its affiliates in the United States and certain other countries. All other trademarks mentioned in this document are the property of their respective owners.

Safe Harbor Statement. This press release contains statements that may constitute forward-looking statements pursuant to the safe harbor provisions of the Private Litigation Reform Act of 1995. These projections or statements are only predictions. Actual events or results may differ materially from those in the projections or other forward-looking statements set forth herein.

Among the important factors or risks that could cause actual events to differ materially from those in the projections or other forward-looking statements are: risks associated with the completion of the acquisition; the impact that failure to do so could have on the realization of the anticipated benefits of the acquisition; risks associated with the successful integration of Vayala Corp.'s business; dependence on new product development; rapid technological and market change; potential fluctuations in quarterly results; acquisition strategy; risks associated with the Internet infrastructure and regulation; volatility of stock price; international operations; financial risk management; and future growth subject to risks.

Readers are referred to the documents filed by Quepasa with the SEC, specifically the most recent reports on Form 10-K, 10-Q and 8-K, each as it may be amended from time to time, which identify important risk factors that could cause actual results to differ from those contained in the forward-looking statements.