<< Previous sql server full text index tool Help | Next sql server full text index tool Help >>

Phrasal search

Phrasal search is a standard kind of search which allows to find documents containing a particular phrase. Despite the triteness of the phrasal search, SearchInform offers a range of interesting settings which make the searching procedure much more convenient and effective.
So to carry out phrasal search you should select the "Search similar" tab in the application's main window tool bar. Let's dwell on each option of this tab.



The first field for entering a text is "It's clear that choosing the correct key words accounts for successful search, so it is not always effective to enter some query literally: a much more effective query is one consisting only of key words (those conveying the core of searched information). Note that if the currently mounted index enables the stemming analisys (view "Indexes manager"), then this function will also be used in the phrasal search. In this case it's not necessary to provide for all the word forms that the key words might take in the searched documents. It will be enough to enter all the words in their initial form.

You should also bear in mind that phrasal search is conducted with the stop words list in active mode ("view Stop words manager"), so even if you enter some words into the search line present in the stop words list (which consists of words with zero informative value), these words will be disregarded by search.

Index field allows to select one of the indexes mounted to the system to further search for key words by this particular index.

The Search button correspondingly launches the process of document search, fitting with all the settings on this tab. This means that it should be pressed only after setting all the options, i.e. in the very end.

The Found words field is filled automatically on the completion of search. It displays all the words from the search line which were found at least in one document. Correspondingly, the words in this list are not duplicated. Double clicking one of the words in this list automatically starts a new search by this word (i.e. the operation is analogous to typing this word in the search line and pressing Search button)

All the following widgets are immediate search options.



The first options block allows to select one of the principles of handling every word in the search line:

"Start from words" - selecting this option means that every single word will be regarded in the following way: if some word was found starting with this key word (i.e. the first symbols of the found words fully coincide with key word), in this case the application will regard these words as equal and will take account of them by search. For example, if the word "auto" was entered into the search line then the words such as "auto", "automobile", "automate" and the like will be considered equal (fitting the query auto).

"Containing words" - this option means that the key words in the search line should be fully contained in the words of the processed document to be regarded as equal. This means that for any two words to be regarded equivalent in this mode, some part of the word viewed by the application should fully correspond with the key word up to the last symbol. In this case it doesn't matter whether the key word corresponds with the beginning, middle or end of the word in the document. For example, the key word "form" will be equivalent to "information", "formula", "performance" etc.

"Terminating on words" - this mode is actually equal to the intersection of the two previous cases, i.e. the words will be considered concordant when the word in the document will end with the key word. In this case the key word and the word fitting with such a criterion will be regarded as equal. For example, in such a mode the search word "plane" will be equal to "aeroplane", "paraplane" etc.

"Conterminous to words" - selecting this mode is basically the intersection of all the previous cases, i.e. the search will be conducted strictly by the key word. The found word should fully correspond with the key word or at least be its stem, when the stemming is used. This mode is selected by the application by default, as most frequently used one. It should also be mentioned that this mode is the fastest one. Using one of the modes described above can considerably lower the speed of search, as in these cases a much larger volume of information should be processed.

Now let's study the following options block:



"All words must present in result" - checking this box means that the document will be considered fitting the query only when it contains all the words from the search line (naturally, except the words from the stop list). At the same time their order, as well as distance between them are of secondary importance.

"Fuzzy search" - checking this box allows to broaden the seek area to some extent. If "Fuzzy search" is checked, SearchInform will look for key words fitting all the set parameters, but taking into consideration possible inaccuracy (for example, spelling of words). The difference of "Fuzzy search" from stemming is that "Fuzzy search" cannot examine all the word forms, it only enables revealing possible minor mistakes or inaccuracies in words, regarding them equal to key words. For example, the document contains a word "appke". It is obviously a mistyped word "apple". "Fuzzy search" will recognize such an inaccuracy and will regard the word as "apple". Stemming on the contrary allows for all the different word forms but it cannot reveal inaccuracies in words, not connected with stemming. To sum up, both the stemming and the "Fuzzy search" separately, and especially their combination can significantly increase the correctness of search. But bear in mind that such analysis can distract a lot of resources, which in its turn will result in slower search.

"Use synonyms" - allows to enable using synonyms by phrasal search. At default this option is active as it makes search much more effective. Detailed information on synonyms is provided in "What you need for high-quality search?" It should only be noted that this function is useful only when there is a substantial list of synonyms (see how to create such a list in "Synonyms manager"), otherwise activating this function makes no sense.

"Phrase search" - if this option is activated, search will be conducted exactly by the phrase, entered into the search line. This search is targeted at determining not only presence of words from the search line in every scanned document, but also their order in regards to each other. It can be said that "Phrase search" searches for a "cite" (expression from the search line), i.e. a phrase in its initial form. However, this opportunity offers a number of additional settings. If the "Word order" checkbox is checked, then the words in the found expression should be in exactly same order as in the "cite". If the box is unchecked, then the found phrase can be different from the "cite" only in word order. The "Word order" option becomes available only after activating the "Phrase search" option. Besides "Word order", "Phrase search" also enables to set maximum permissible number of unnecessary in-between words in the "cite". This allows to carry out search with less rigid conditions. The setting is done with the help of a toddler, and the permissible number of in-between words is displayed in form of "Number of in-between words: X". Correspondingly, the number of unnecessary words between each pair of key words can be less or equal to the fixed value. If the number exceeds the value, the phrase is considered irrelevant.



Taking into consideration the fact that the number of documents found by some query can be really huge, it would be logical to limit the maximum number of shown results in order not to waste time on processing excessive data. Moreover, the last results are usually of very low relevance to the search query, as they are all sorted by relevance. To limit the number of shown documents, you should set the desired number in the "Show first X result" field.


Often on completing the search, we need to make it more specific, applying new search criteria, but this time to search only through the obtained results. To do this, utilize the "Search through current results" option.
Here is a small example of how to practically use this function. For example, we have completed some search by a certain set of key words, with the "All words must present in result" box unchecked. This has resulted in a too long list of results. Now we have decided that a particular word from the search line should be present in the target document without fail. This can be implemented by activating the "Search through current results" option. Double-clicking the necessary word in the "Found words" field will display only those of the documents from previous query that contain the selected word.

Unique similar search technology in our sql server full text index tool utility At present most of the potential customers simply have no clue that there is a sql server full text index tool tool for searching similar documents. The problems facing a corporate user can be solved nice and easy. And the name of the game is the SoftInform sql server full text index tool technologies. It is important, first of all, to inform the corporate user on SoftInform sql server full text index tool solution, its originality and strengths as compared to sql server full text index tool competitors, and second of all, to provide turn-key projects (examples) that would demonstrate the ease with which any challenge of corporate search and structuring information is solved by means of our sql server full text index tool technology.

<< Previous sql server full text index tool Help | Next sql server full text index tool Help >>

Our full text search engine system allows you to search web site

When you submit your Site for indexing, our full text search engine spider will "crawl" your site, traversing each and every link. Each page is "read" and a "search index" is created. This index is stored on our servers. The HTML code you insert into your page produces a search button and a query text box. It also contains a link back to our full text search engine tool. When a search is requested, our computers search your index and a results page is created. This page is returned to the searcher's browser. Everything happens here. You see the results there.