Pursue better search results

Intermediary transaction SEO diagnosis Taobao guest Cloud host technology Hall

Moore's Law describes the fact that an integrated circuit semiconductor can double over a number of years. Similarly, the World Wide Web is growing at a rate of one times every 9 months, and will continue to grow exponentially. Thankfully, Moore's law and its unnamed reasoning continue to increase the speed and overhead of increasing network load. On the other hand, finding what you need from the Web or data warehouse is not progressing fast, although retrieving related results is growing at a megabyte or even 10 times times faster.

It is clear that the development of high quality search results has failed to go hand in hand with Moore's law, which doubled over the years. In fact, the law of search results can be described in terms of the increasing size and complexity of the data.

The May 2004 Nielsen Norman Group released its web Usability survey, which highlighted that search has become one of the great hurdles on the road to Web application success. According to the survey, access to search engines has become the first step in the 88% Web session, while the average user accesses 3.2 sites per session (except access to search engines).

For search success, all users, including those who are occasionally or experienced, are only 42% percent satisfied with the search results. Although this number can reach 50% for experienced users alone, it still means failure.

The worst search experience often occurs when you use a corporate site search instead of using a giant search engine. Jakob Nielsen, an expert in search usability and head of Nielsen Norman Group, describes the search function of the intranet as "worthless." Many users even try to avoid companies approving the development of intranet search engine projects, preferring to use web search engines to find information about their company outside the firewall.

Part of the problem is that search is essentially an input-output stream. About 60% of the surveyed users typed only one word at the beginning of the search, while the other 20% typed two words. Only 1% of the respondents used advanced search capabilities, while only 3% of users were further optimized for search using quotes or other query syntax. In addition, the survey showed that the first link in the search results page received 51% hits, and the second one was 16%. The same result is derived from the search engine Company's database, and there is another law on search: Don't expect users to use more tools than the search engine provides to get information.

Nielsen recommends some basic strategies to modify the user's query action, such as a simple visual search box, which has at least 27 letters wide to encourage users to make multiple-word queries, with spell checking and the ability to manually modify the query entries.

"You can add proactive judgments to search engines by extracting the first 1000 search entries and judging where the important places to find these words or phrases are usually," he said. Nielsen said. "You can also check the search logs to see which items are most commonly used and which words are used to search for them, and then add synonyms to the search engine." ”

In addition, Nielsen mentions that using appropriate content elements such as page titles, headlines, and summaries can help improve search results. Another search law is drawn from these steps: "Relying on automated queries alone will not result in satisfactory results." ”

At the same time, enterprise network data is organized and unstructured (such as e-mail, Office documents, Web pages, audio files, and so on), and the data is in a separate state and is not ready for retrieval. Another search law arises: You can't look for what you want in something you don't know.

With the results of the Nielson Norman Group and the information on usage patterns obtained through the survey, search engines have to take steps to make themselves smarter, with a more accurate and comprehensive search result for most users. Most people do not have to struggle with advanced search techniques to overcome the limitations of existing searches. At the very least, most companies need to check their search mechanisms, increase their input into the classification of raw data, and adopt a professional search engine with dedicated team maintenance and upgrades.

Companies such as Autonomy, FAST, Google, Northern Light, Verity and Vivisimo offer enterprise search engines using a variety of technical means. Google has its own "PageRank" and text-matching techniques. Autonomy uses such as concept matching (concept matching) technology, Bayesian decision rule inference technique, etc.

Specialized search engines focus on specific areas, such as Globalspec for engineering information, which can provide better results than a strong search engine in general. As IBM is currently studying the Webfountain, it runs thousands of programs to constantly retrieve information and categorize content, while using natural language analysis to provide contextual reference. Webfountain is designed for highly complex queries, such as tracking the visibility of a company or product.

However, do not expect a major breakthrough in the search engine area recently. "Today's search engines are like 8-track tapes from the History of music," said Gary Flake, chief scientist at Yahoo's search lab. "At least for now, this gives rise to the ultimate law of search results--take the liberty of quoting Shakespeare's lines:" All day long, and once possessed, it is found that the effort in pursuit exceeds the value gained.