Advanced Content Search
Powerful engine equipped to "grow up" and "be found"
Neighborhood search such as position consideration search and Boolean expression search are also available, which is known as the most powerful Japanese search engine. Function pages, such as search results, that used to generate and analyze statistical results in order to easily match search content requested by visitors, are also provided.
Since there is learning software, searched content will be corrected with time. Moreover, searched keywords can be stored according to user properties, each of the searched keywords can be separately distributed. For instance, a female visitor that used cosmetic products as her search keyword. Essential content will be displayed and controlled.
In the case where information could not be found by site visitors, in-site search will be used. However, if essential information is not available, the site may provide mismatched information. The search engine installed in HeartCore does not only avoid search result mismatching, it also benefits from raising the search efficiency, upgrading usability, and increasing conversion at the same time.
Pinpoint search
Synonymous search (natural sentence search, ambiguous search)
QS hybrid method (N-gram based statistical processing and morphological analysis)
| Search method | N-gram method | Morphological analysis | QS hybrid method |
|---|---|---|---|
| System dictionary | Not required | Required | Required (not necessarily) |
| Index format | N-gram inverted file: Full text index |
Word inverted file: Word index |
N-gram inverted file+ word inverted file |
| Pros | Arbitrary word searching No retrieval omission |
Noises are not likely to interfere with search results | -Arbitrary word search -No retrieval omission -Noises are not likely to interfere with results |
| Cons | Potential noises interfered with search results -Search for‘Kyoto' and hits‘Tokyo’ -Search for‘Ski’and hits‘whisky' |
Retrieval omissions →Unable to recover Unable to retrieve buzz words →Dictionary maintenance is needed, and index recreation is required -Model number partial match -Search‘intra' and will not hit 'intranet' |
-Big index size (1.1 times bigger than original) -Index updating time remains the same (but high CPU and memory load) *Dictionary maintenance is not required with regeneration of index, in order to cover retrieval omissions by N-gram base |
Ambiguous search and integrations with other dictionaries
Suggestion
Keyword search histories will be used to suggest auto-complete keywords following by the input keyword.
Keywords searched in the past will be a keyword candidate automatically, and not only for list up:
-Administrator can set suggested keyword on purpose as desired
-Administrator can appoint suggested forbidden keyword and stop display on suggest column
*It is possible to appoint by regular expression of the forbidden suggested keyword in front/rear/middle agreement.
General score: convenient scoring feature
-Indicator, such as score for a full article, updated time of data (freshness), or access number is possible to add generally ranking indication.
-For EC site operators, flexible search target rank tuning will be provided.
Access ranking
In the sequence of search results, the order of consideration of the number of accesses of each record is enabled.
Value keywords in specific fields
- Add priority levels from 1(lowest) to 5(highest) for fields
- Tune rankings for fields
For example, the product database has columns named "title" and "details". Product A contains a word “低燃費” once in the title field and twice in the details field, while Product B does not contain the word “低燃費” in the title field and happens to contain it five times in the details field.
When users search for “低燃費”, Product A is most likely the one they are looking for. With the QuickSolution standard ranking method, however, Product B will be ranked higher than Product A as the keyword appears more frequent in Product B.
Adding priority to the specific fields, it will be considered more relevant that the keywords appear in the fields specified.
High speed sorting by given fields
High speed processing even with over one-hundred thousand hits
Multiplex sorting by any combination of the attribute fields
Smart link
Simple clicks for narrowing search results by keywords and/or related words
Even with a huge amount of data, no special operation is required to retrieve advanced search results.
Category count
Similar document collection: Gather up similar products
Various reports export
This search in site equipped as default functionality in HeartCore, is run by a big data era search engine, "QuickSolution" developed and provided by Sumitomo Electric Information Systems Co., Ltd.
* The search engine, "QuickSolution" is a registered trademark of Sumitomo Electric Information Systems Co., Ltd.
Please refer to the following product website for more information on "QuickSolution".
http://www.sei-info.co.jp/quicksolution/index.html
