2. SPARK: Adapting Keyword Query to Semantic Search (ISWC2007)
   http://iswc2007.semanticweb.org/papers/687.pdf
这篇是上海交通大学的ISWC07的论文，对上一篇中第(3)步构建formal query的步骤采用先分割query set，再使用minimum spanning tree算法构建 query graph，最后利用一些概率的方法来完成ranking query，得到Sparql查询语句。

3. Ontology-based Interpretation of Keywords for Semantic Search (ISWC2007)
   http://iswc2007.semanticweb.org/papers/519.pdf
这篇论文主要对第(3)步构建formal query的步骤提出了如何connect第(2)步中得到的entities来构建query query graph的算法，ranking query采用简单的path lenth为标准。

4. Q2Semantic: A Lightweight Keyword Interface to Semantic Search (ESWC2008)
   www.aifb.uni-karlsruhe.de/WBS/dtr/papers/eswc08_q2semantic.pdf
这篇是上海交通大学的ESWC08的论文，是对之前方法改良较大的一篇，最重要的改进是引入了clustered RACK graph(a summarz of the original RDF graph)的概念和方法，比如把original RDF graph中所有的instances都map进其相应的concept节点等等，这样大大减小了RDF graph的size，使得query construction和ranking更加快速，另外一个好的改进在于:ranking query不仅考虑了keywords matching relevance，path length，还考虑类似于Google PageRank的Importance of Edges and Nodes。
但同时也引入了一些问题，由于构建query时没有考虑instance和instance之间的连接(instance都map进相应的concept)，所以有可能得到的formal query(Sparql查询)查询不到任何结果。

个人的理解，可能有些地方不准确。

当然还有其他的一些，在这上面有一些相关的研究动态以及应用系统体验！

效率记得好像版主有个帖子比较过。

至于它的推理性能，因为没有使用过，因此也不便评论。不过若如楼上所言，应该属于整合的推理方式，至于是否节约推理时间，个人认为最好做一个测试来进行比较。不知楼上以为如何？
我可能没有表述清楚！他们提出这种方法适用于基于本体模式映射的集成环境，此处应该不是推理性能的问题！如果你阅读了DERI的相关delivery，会发现此种方法是一种技术上的创新！

欢迎其他的同道来此分享！把自己的想法用几句话说明白也是一种写作的锻炼嘛！

对于“本体查询语言”其实我还没有进行过深入了解，不过据我所知目前推理引擎对某一种或几种本体查询语言提供了支持。我感觉它们之间的关系类似理论和实践的关系：推理引擎提出了一种语义推理的思想和方法，而操作则需要通过具体的查询语言和本体库作为“工具”。或许这种理解有失偏颇，不过一个只能想不能做的“推理引擎”是没什么用处的。

Humphrey的想法和我的有些不同，接下来探讨一下。1）不通过推理引擎，SPARQL是否可以执行呢？因为据我所了解的情况，所有的推理引擎均不支持完整的SPARQL语法，但网上有一些SPARQL在线执行的例子却可以完整的支持。那么是否可以得出结论，检索语言和推理引擎并不需要配套使用。2）通过推理引擎，SPARQL获得了那方面的支持那？换句话说，和不加载推理引擎所完成的检索在结果集上有什么区别那？
感觉还是有点糊涂！！大家支支招！

第一个观点我很赞同，后来考虑了一下，的确应该如Humphrey所说；
但是后面一个观点我有点不同的想法，如果仅仅是出于优化、使用习惯方面来考虑检索，那么本体所体现出来的“显式的语义表达与描述”去哪里了呢？跟一般的作用于数据库模式上的SQL又有何区别呢？既然叫做“推理引擎”，总得做点“顾名思义”的工作吧！！

和Humphrey同志的研究方向可能有差别，我现在比较关注基于模式方面的，而Humphrey则是类似于现在的Google搜索引擎类的。因此，研究的重点也会有所区别。模式集成往往是有领域限制的，因此形成共识就会有一定的基础；但是搜索引擎面向整个Web，因此想要做到粒度够精细的程度，不太现实。因此现在为止通用的标准也只有DC、FOAF等寥寥几个。如果仅仅是通过一个同义词库来实现，是否也可以说是语义检索呢？呵呵！定义太混乱！！

——————————————————
关系数据库查询和语义查询之间的关联和转换不知道有没有其他人在做？大家一起讨论讨论啊

Pellet supports SPARQL, which will allow you to make ontology queries.
However, SPARQL is an RDF-based query language and has no understanding
of OWL. It will, however, work with OWL if you have very simple queries.
I'm not sure how it will interact with SWRL though.

SPARQL can express RDF queries and will work with OWL in some
circumstances. However, its lack of understanding of OWL's semantics can
make it incredibly difficult to write many OWL queries.

对于rULE有SQWRL来处理：

1. Understands OWL with RuleML/SWRL
2. Handels queries in form of rules

o'k！看来得扩展一下思路了！
http://protege.cim3.net/cgi-bin/wiki.pl?SQWRL

正在阅读中！后续再聊！

http://www.devx.com/semantic/Article/38595/1954?pf=true

Semantic Search Arrives at the Web
The current generation of search engines is severely limited in its understanding of the user's intent and the web's content. Find out how semantic search can address these limitations.  

by Peter Mika

很好的直觉.数据挖掘的算法也是为了发现关系.而语义网则是显示表示和处理关系.所以数据挖掘的确能达到语义搜索的效果.但精确度肯定是不够的.
数据挖掘一般都需要有一个数据模型.所以如果语义网的确定模型应用到数据挖掘中可能会有意想不到的效果.
大家共同研究

http://semantic-mediawiki.org/wiki/Help:Semantic_search

不能算吧。

Semantic search是一个很含糊的名词，包括Natural language query, Semantic Web search等。但是无论如何，他是基于搜索的，所以我们不能撇开Lucene等搜索框架而重头设计一套东西。需要了解现有框架或者系统的不足，以及考虑哪些地方可以扩展。

to richardjya，语义搜索包含很多内容，即使是semantic web search也有很多步骤运用不同的语义技术，因此说语义搜索走不下去未免过于绝对。同时我认为现阶段的semantic search肯定不是基于NLP的，而应该是考虑如何最大限度地利用meta data，那么就涉及到如何exploit或者leverage这些graph-based data，会有新的ranking问题，有新的更scalable的图遍历算法等产生，甚至对search的整体infrastructure进行修改。拿google作例子，可能会对GFS, Bigtable, Map-reduce等进行必要的修改和扩展。所以我觉得结构发现和匹配是关键。你不可能让用户是指定SPARQL或者指定其中包含的结构，但是不代表semantic search不要利用这些资源或者信息。

基于关键字的语义搜索：
SemSearch: A Search Engine for the Semantic Web. EKAW 2006
A Relation-based Search Engine in Semantic Web. TKDE 2007
SPARK: Adapting Keyword Query to Semantic Search. ISWC 2007
Ontology-based Interpretation of Keywords for Semantic Search. ISWC 2007
Q2Semantic: A Lightweight Keyword Interface to Semantic Search. ESWC 2008
Top-k Exploration of Query Candidates for Efficient Keyword Search on Graph-Shaped (RDF) Data. ICDE 2009.

关于自然语言查询接口，你可以关注PowerAqua (KMi的工作)以及Controlled Natural Language (University of Zurich的工作)，具体的paper列表我就不列了。

另外和你说的相关的还有faceted browsing或者叫exploatory search，最简单的就是很多e-business的购物网站，提供各种product的属性等来约束搜索结果，同时，query relaxing是另外一个相关的topic，他的想法是根据用户的preference或者当前的search context对于某些搜索条件添加不同的权重，或放宽搜索约束条件。

希望这些简单的解释对你有帮助。

Top 10 Semantic Web Products of 2008
Written by Richard MacManus / December 2, 2008 9:57 AM

In 2008 we saw the Semantic Web gain traction, giving us plenty of choice when selecting the 10 best Semantic Web products of 2008.
This is the first in a series of posts we'll publish over December, listing our choices for the top web products of the year. Then at the end of December, we'll post a Top 100 list - which we'll be promoting over 2009 and opening up at some point for public voting. Without further ado, let's jump into the top 10 Semantic Web products of 2008.
Earlier this month we posted an update to 10 Semantic Web applications that we have been tracking for a year now. Some of those make this list, as well as some from our follow-up post 10 More Semantic Apps to Watch. We also have a couple of other products in this list, which for one reason or another didn't get mentioned in our watch-lists.
You may disagree with our selections, so do tell us in the comments what you think.
Note: the products listed below are in no particular order
1. Yahoo! SearchMonkey
In May this year Yahoo! launched an open developer platform for search called SearchMonkey. Yahoo hasn't had the happiest of years, but its willingness to innovate in search is to be commended. As we reported at the Web 2.0 Expo in April, SearchMonkey is a component of a major overhaul at Yahoo! across all of its properties to "rewire" for the social graph and data portability. SearchMonkey allows developers to build applications on top of Yahoo! search, including allowing site owners to share structured data with Yahoo!, using semantic markup (microformats, RDF), standardized XML feeds, APIs (OpenSearch or other web services), and page extraction.
We think this is the best use of Semantic Web by an Internet bigco this year. So for that reason SearchMonkey makes our top 10 list. Related: The Story of SearchMonkey.
2. Powerset (acquired by Microsoft in '08)
Powerset (see our initial coverage here and here) is a natural language search engine. It's fair to say that Powerset has had a great 2008, having been acquired by Microsoft in July this year.
At the time of the acquisition, Powerset said that it needed a bigger partner to expand its product beyond its current state of only searching Wikipedia - something we had speculated about when the rumors of the acquisition first appeared. In its own statement, Microsoft stressed how useful Powerset's technology will be for improving Microsoft's own search products and to "take Search to the next level." In our analysis of the deal, we noted that it was a "bold play requiring exact execution" by Microsoft.
3. Open Calais (Thomson Reuters)
At the end of 2007, ClearForest had been recently acquired by Reuters and at that point it had a Web Service and a Firefox extension. What a change a year brings! ClearForest went on to release Calais, a toolkit of products that enable users to incorporate semantic functionality within their blog, content management system, website or application.
Since launching the Open Calais API early this year, over 6,000 developers have registered with it and the service is doing more than 1 million transactions a day. Version 3.0 was released earlier this month and version 4 is expected by January 09.
4. Dapper MashupAds
In November we wrote about the recent improvement in Dapper MashupAds, a product we first spotted over a year ago. The idea is that publishers can tell Dapper: this is the place on my web page where the title of a movie will appear, now serve up a banner ad that's related to whatever movie this page happens to be about. That could be movies, books, travel destinations - anything. We remarked that the UI for this has grown much more sophisticated in the past year.
The company believes that its new ad network will provide monetary incentive for publishers to have their websites marked up semantically. We think this has plenty of promise, so it makes our year-end list.
5. Hakia
Hakia is a search engine focusing on natural language processing methods to try and deliver 'meaningful' search results. Hakia attempts to analyze the concept of a search query, in particular by doing sentence analysis. Over the past year Hakia has been busy extending its reach - licensing its proprietary OntoSem technology to other companies in March and announcing a Semantic API in June. It was also one of the first companies to utilize Yahoo! BOSS, by integrating their semantic parsing with the Yahoo! search index.
We think Hakia has made good progress getting its technology into the hands of third parties and making use of Yahoo's broader index, so for that reason it's among our top 10 for the year.
6. TripIt
Tripit is an app that manages your travel planning. With TripIt, you forward incoming bookings to plans@tripit.com and the system manages the rest.
Over the past year TripIt has continued to iterate on its feature set - introducing LinkedIn integration, better mobile functionality, more social networking features, and other goodies. In short, it's user experience continues to rock!
7. BooRah
BooRah is a restaurant review site that we first reviewed earlier this year and has come on in leaps and bounds over 2008. BooRah uses semantic analysis and natural language processing to aggregate reviews from food blogs. Because of this, BooRah can recognize praise and criticism in these reviews and then rates restaurants accordingly. BooRah also gathers reviews from Citysearch, Tripadvisor and other large review sites.
BooRah also announced last month the availability of an API that will allow other web sites and businesses to offer online reviews and ratings from BooRah to their customers. The API will surface most of BooRah's data about a given restaurant, including ratings, menus, discounts, and coupons.
8. AdaptiveBlue
Disclosure: AdaptiveBlue's founder Alex Iskold is a feature writer at RWW.
AdaptiveBlue are makers of the Firefox plugin, BlueOrganizer. As we wrote in January this year, the basic idea behind BlueOrganizer is that it gives you added information about webpages you visit and offers useful links based on the subject matter.
Over the past year the company has been working on a new product, called Glue. Launched last month, Glue is a more social networking oriented version of BlueOrganizer - it connects you to your friends based around things like books, music, movies, stars, artists, stocks, wine, restaurants, and more. We think the company has diversified smartly in 2008, by integrating social networking and mobile functionality into its products.
9. Zemanta
Zemanta is a blogging tool which harnesses semantic technology to add relevant content to your posts. While it didn't make either of our 'Semantic Apps to Watch' lists in November, a number of commenters pointed it out as something they use. In September we covered a major upgrade to Zemanta's service, allowing users to specify the sources they want to see in the suggestions list that Zemanta provides. Users can now incorporate their own social networks, RSS feeds, and photos into their blog posts. As we noted, this makes Zemanta a lot more appealing to established bloggers who are in less need of suggestions and more in need of automation.
Zemanta's API is also being used by startups, including semantic bookmarking service Faviki - which we mentioned in our second Watch-list. So all up, we think Zemanta has done enough this year to be included in our top 10 list.
10. UpTake
Semantic search startup UpTake (formerly Kango) aims to make the process of booking travel online easier. In our review in May, we explained that UpTake is a vertical search engine that has assembled what it says is the largest database of US hotels and activities - over 400,000 of them - from more than 1,000 different travel sites. Using a top-down approach, UpTake looks at its database of over 20 million reviews, opinions, and descriptions of hotels and activities in the US and semantically extracts information about those destinations.
And now please let us know in the comments what you think of our selections. Do you think we've picked the best 10 Semantic Web products of the year?