Mailing-List: contact solr-user-help@lucene.apache.org; run by ezmlm
Precedence: bulk
Reply-To: solr-user@lucene.apache.org
Received-SPF: pass (athena.apache.org: domain of jtao@webmd.net designates
 157.56.110.103 as permitted sender)
From: "Tao, Jing" <jtao@webmd.net>
To: "solr-user@lucene.apache.org" <solr-user@lucene.apache.org>
Subject: protected phrases - possible?
Thread-Topic: protected phrases - possible?
Thread-Index: AdBrDkidrgIzKjKDRgmWgt22TxwX1g==
Date: Mon, 30 Mar 2015 17:48:16 +0000
Message-ID: 
 <6DFE5FE777890F4786F01470C2BFDFD74224E2A8@EXMBX01L-CRP-03.webmdhealth.net>
Accept-Language: en-US
Content-Language: en-US
Content-Type: multipart/alternative;
	boundary="_000_6DFE5FE777890F4786F01470C2BFDFD74224E2A8EXMBX01LCRP03we_"
MIME-Version: 1.0
Received-SPF: Pass (protection.outlook.com: domain of webmd.net designates
 207.138.251.38 as permitted sender) receiver=protection.outlook.com;
 client-ip=207.138.251.38; helo=exht01l-crp-03.webmdhealth.net;
X-MS-Exchange-CrossTenant-OriginalArrivalTime: 30 Mar 2015 17:48:18.2354
 (UTC)
X-MS-Exchange-CrossTenant-Id: 8423b0e1-4e0c-47b0-8280-02b47a7165e1
X-MS-Exchange-CrossTenant-OriginalAttributedTenantConnectingIp: 
 TenantId=8423b0e1-4e0c-47b0-8280-02b47a7165e1;Ip=[207.138.251.38];Helo=[exht01l-crp-03.webmdhealth.net]
X-MS-Exchange-CrossTenant-FromEntityHeader: HybridOnPrem
X-MS-Exchange-Transport-CrossTenantHeadersStamped: BY2PR08MB189
X-Virus-Checked: Checked by ClamAV on apache.org

--_000_6DFE5FE777890F4786F01470C2BFDFD74224E2A8EXMBX01LCRP03we_
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: quoted-printable

Hi,

The way our collection is setup, searches for "breast cancer" are returning=
 results for ovarian cancer, or anything that contains either "breast" or "=
cancer".  The reason is, we are searching across multiple fields.  Even tho=
ugh I have set a "mm" value so that if less than 3 terms, ALL terms much ma=
tch...SOLR considers it all matched even though "breast" was in the title a=
nd "cancer" is in the description.

Is there a way to protect certain phrases so that they will not be tokenize=
d?  I tried using CommonGramsFilterFactory, but having "breast cancer" in t=
he word list did not seem to do anything.  I'm guessing it's because the fi=
eld is tokenized first, so nothing would match that phrase.  If I put "brea=
st" and "cancer" as separate entries in the word list, I end up with too ma=
ny unnecessary shingles, and "breast" and "cancer" are still two of the fin=
al terms.

I have a feeling CommonGramsFilterFactory is not the right way to handle th=
is.  What are other options?  Is it better to put all fields in one field, =
apply mm, and proximity boost?

Thanks!
Jing

--_000_6DFE5FE777890F4786F01470C2BFDFD74224E2A8EXMBX01LCRP03we_--