News
By Sandra Rosenzweig
Searching by Concept
Whenever you see the words practice management, the word search can't be far behind. Or, to put it another way, improvements to search engines mean improvements to practice-management software. That's how the practice-managementprograms from LexisNexis (Total Practice Advantage, http://law. lexisnexis.com/total-practice-advantage) and Thomson-Elite (ProLaw, www.thomsonelite.com/solutions/product-fam/prolaw/ index.asp) get their clout. Both are built around and integrated into their parent companies' search methods. Other programs integrate with third-party engines, most notably World Software's Worldox, which provides a simple interface for creating searches (www.worldox.com).
So far, however, search technologies mainly use full-text searching, but the feeble results miss all sorts of relevant hits. Some search programs rely on your previous 100 searches to help formulate the queries' results. Some?mostly desktop searches?rely on pre-indexed indices. Others use synonym lists to get you close to your goal of finding, say, who has done research on medical marijuana by also automatically looking for hashish, pot, Mary Jane, weed, and grass. Both full-text and synonym techniques may get you closer to your goal, but those little search bots are just as likely to miss the point of the query.
If you are using a search engine purchased by your firm, and if it isn't ISYS (www.isys-search.com) or Google, you can improve the accuracy and relevance of your hits noticeably by constructing a customized firmwide taxonomy based on tags (formally called metatags, or preset categories). In a long series of meetings, the firm's lawyers (not clerks, paralegals, or assistants) agree on the high-level groupings, subtopics, and appropriate tags within each rubric. Or they agree to borrow headings offered by West or Lexis as the armature on which to hang the firm's tags. So far so good, but wouldn't it be a smashing idea to search only on, say, the contracts that have been used by lawyers in the firm and accepted by the court? Then, you have to hold more meetings to set up appropriate tags for used and failed, used and accepted, never used?that sort of thing. And then comes the meeting at which the firm's lawyers learn that every time they create, read, or edit a document, they must apply tags?and that the lawyers themselves have to do the tagging (because the clerks, paralegals, and assistants don't know the practice of law well enough to understand the tags). Even with a structured checklist of tags, the tagging process will require the lawyers' attention, bore them to tears, and take up their time. And suddenly, you find yourself alone in the room.
For the past few years, several search companies have claimed they already base their programs on concept searching, but in truth the technique is in its infancy. Even Google Labs is working on it. Google Experimental Search (www.google.com/ experimental) produces maps and timelines. Google Suggest (www.google.com/webhp?complete=1&hl=en) offers you a list of contexts for your search query, so that your search for COBRA (the Consolidated Omnibus Budget Reconciliation Act of 1985), won't bring you any cobra snake hits.
Until fully fledged concept searching arrives, though, try this: Go to Google Sets (labs.google.com/sets) and enter into the program three to five keywords that you consider relevant to your search. Then click on the Small Set (15 items or fewer) button, and Google Sets will return several other, related topics. (If the Small Set isn't on the mark, try making a Large Set and see if that is any more relevant.) For example, I entered Consolidated Omnibus Budget Reconciliation Act, Social Security, and Medicare, and the program sent back Medicaid, food stamps, immigration, education, healthcare, President Bush, unemployment benefits, veterans, politics, and a few more. I took those hits and created a few queries: Medicare and Veterans; Healthcare and President Bush; Social Security, veterans, and politics. For veterans and politics, I got more than 3.1 million hits. Then within those results (a link at the bottom of each Google hits page), I searched on one of my keywords, Agent Orange, and got about 750,000 hits. When I added healthcare, I got down to about 350,000 hits. I would have refined the search yet again but for the coincidence that the article of my dreams was the second hit down: "Offshore Vietnam Veterans Finally Receive Compensation and Healthcare," by Morgan Barker. I hadn't supplied an author's name, nor a precise title. Yes, it would have been simpler to just type the keywords into the search box, without any additional steps. However, I didn't know some of the most useful keywords?until I went through those context choices. If I already had those terms, I would just have put a string of keywords into an ordinary Web search engine and received back some relevant links. But I needed that article to guide me toward my keywords, so the extra steps really paid off.
Searching by Concept
Whenever you see the words practice management, the word search can't be far behind. Or, to put it another way, improvements to search engines mean improvements to practice-management software. That's how the practice-managementprograms from LexisNexis (Total Practice Advantage, http://law. lexisnexis.com/total-practice-advantage) and Thomson-Elite (ProLaw, www.thomsonelite.com/solutions/product-fam/prolaw/ index.asp) get their clout. Both are built around and integrated into their parent companies' search methods. Other programs integrate with third-party engines, most notably World Software's Worldox, which provides a simple interface for creating searches (www.worldox.com).
So far, however, search technologies mainly use full-text searching, but the feeble results miss all sorts of relevant hits. Some search programs rely on your previous 100 searches to help formulate the queries' results. Some?mostly desktop searches?rely on pre-indexed indices. Others use synonym lists to get you close to your goal of finding, say, who has done research on medical marijuana by also automatically looking for hashish, pot, Mary Jane, weed, and grass. Both full-text and synonym techniques may get you closer to your goal, but those little search bots are just as likely to miss the point of the query.
If you are using a search engine purchased by your firm, and if it isn't ISYS (www.isys-search.com) or Google, you can improve the accuracy and relevance of your hits noticeably by constructing a customized firmwide taxonomy based on tags (formally called metatags, or preset categories). In a long series of meetings, the firm's lawyers (not clerks, paralegals, or assistants) agree on the high-level groupings, subtopics, and appropriate tags within each rubric. Or they agree to borrow headings offered by West or Lexis as the armature on which to hang the firm's tags. So far so good, but wouldn't it be a smashing idea to search only on, say, the contracts that have been used by lawyers in the firm and accepted by the court? Then, you have to hold more meetings to set up appropriate tags for used and failed, used and accepted, never used?that sort of thing. And then comes the meeting at which the firm's lawyers learn that every time they create, read, or edit a document, they must apply tags?and that the lawyers themselves have to do the tagging (because the clerks, paralegals, and assistants don't know the practice of law well enough to understand the tags). Even with a structured checklist of tags, the tagging process will require the lawyers' attention, bore them to tears, and take up their time. And suddenly, you find yourself alone in the room.
For the past few years, several search companies have claimed they already base their programs on concept searching, but in truth the technique is in its infancy. Even Google Labs is working on it. Google Experimental Search (www.google.com/ experimental) produces maps and timelines. Google Suggest (www.google.com/webhp?complete=1&hl=en) offers you a list of contexts for your search query, so that your search for COBRA (the Consolidated Omnibus Budget Reconciliation Act of 1985), won't bring you any cobra snake hits.
Until fully fledged concept searching arrives, though, try this: Go to Google Sets (labs.google.com/sets) and enter into the program three to five keywords that you consider relevant to your search. Then click on the Small Set (15 items or fewer) button, and Google Sets will return several other, related topics. (If the Small Set isn't on the mark, try making a Large Set and see if that is any more relevant.) For example, I entered Consolidated Omnibus Budget Reconciliation Act, Social Security, and Medicare, and the program sent back Medicaid, food stamps, immigration, education, healthcare, President Bush, unemployment benefits, veterans, politics, and a few more. I took those hits and created a few queries: Medicare and Veterans; Healthcare and President Bush; Social Security, veterans, and politics. For veterans and politics, I got more than 3.1 million hits. Then within those results (a link at the bottom of each Google hits page), I searched on one of my keywords, Agent Orange, and got about 750,000 hits. When I added healthcare, I got down to about 350,000 hits. I would have refined the search yet again but for the coincidence that the article of my dreams was the second hit down: "Offshore Vietnam Veterans Finally Receive Compensation and Healthcare," by Morgan Barker. I hadn't supplied an author's name, nor a precise title. Yes, it would have been simpler to just type the keywords into the search box, without any additional steps. However, I didn't know some of the most useful keywords?until I went through those context choices. If I already had those terms, I would just have put a string of keywords into an ordinary Web search engine and received back some relevant links. But I needed that article to guide me toward my keywords, so the extra steps really paid off.
#335232
Megan Kinneyn
Daily Journal Staff Writer
For reprint rights or to order a copy of your photo:
Email
Jeremy_Ellis@dailyjournal.com
for prices.
Direct dial: 213-229-5424
Send a letter to the editor:
Email: letters@dailyjournal.com



