promoted_content.csv provides additional details about the ad. There is also the "document_id" parameter. Does that document_id refer to details about the ad itself (describes the ad) or does it refer to a document where the ad is shown?
\n\nIt's somewhat ambiguous to me what a document refers to. It definitely refers to what a user is visiting from the location bar in the browser, but it's not quite clear whether the same document set also refers to the 'ads' (which are essentialy referral documents, but paid for, so not ads in the typical sense), or whether it describes the documents to which the ads were assigned.
\n\nCould someone clarify this please?
", "messages": [ { "id": "138515", "postDate": "10/09/2016 09:46:58", "content": "promoted_content.csv provides additional details about the ad. There is also the "document_id" parameter. Does that document_id refer to details about the ad itself (describes the ad) or does it refer to a document where the ad is shown?
\n\nIt's somewhat ambiguous to me what a document refers to. It definitely refers to what a user is visiting from the location bar in the browser, but it's not quite clear whether the same document set also refers to the 'ads' (which are essentialy referral documents, but paid for, so not ads in the typical sense), or whether it describes the documents to which the ads were assigned.
\n\nCould someone clarify this please?
", "rawMarkdown": "promoted_content.csv provides additional details about the ad. There is also the \"document_id\" parameter. Does that document_id refer to details about the ad itself (describes the ad) or does it refer to a document where the ad is shown? \r\n\r\nIt's somewhat ambiguous to me what a document refers to. It definitely refers to what a user is visiting from the location bar in the browser, but it's not quite clear whether the same document set also refers to the 'ads' (which are essentialy referral documents, but paid for, so not ads in the typical sense), or whether it describes the documents to which the ads were assigned.\r\n\r\nCould someone clarify this please?", "votes": null }, { "id": "138523", "postDate": "10/09/2016 10:04:29", "content": "https://www.kaggle.com/c/outbrain-click-prediction/forums/t/24154/redundant-field-in-database
\n\n\n\n\nThe document_id associated with each ad_id in the promoted content is the document to which the ad refers (i.e. you would use this table in order to find the topics, categories, or entities referred to in an ad).
\n
I was also a little bit confused with promoted_content.document_id
\n\nGood luck!
", "rawMarkdown": "https://www.kaggle.com/c/outbrain-click-prediction/forums/t/24154/redundant-field-in-database\r\n\r\n> The document_id associated with each ad_id in the promoted content is the document to which the ad refers (i.e. you would use this table in order to find the topics, categories, or entities referred to in an ad).\r\n\r\nI was also a little bit confused with promoted_content.document_id\r\n\r\nGood luck!", "votes": null }, { "id": "138800", "postDate": "10/10/2016 22:31:30", "content": "\n\n\nIt's somewhat ambiguous to me what a document refers to.
\n
Document = Web Page. Here is the hierarchy:
\n\nHope this is clear
", "rawMarkdown": "> It's somewhat ambiguous to me what a document refers to.\r\n\r\nDocument = Web Page. Here is the hierarchy:\r\n\r\n1. A document is the largest container \r\n2. Within a document, there is a main content \r\n3. The document will host a display A display is a set of ads - this is randomized every time you visit that same page \r\n4. Each ad may or may not come from the same campaign, or advertiser \r\n5. There may be only 4 ads limited per display - could be more, but I have actually seen only 4 at a time\r\n6. A user visiting a document will generate a page view log\r\n7. A user clicking on a ad will generate a click log\r\n8. Every time the user visits a page, an event is generated and logged\r\n\r\nHope this is clear", "votes": null }, { "id": "152292", "postDate": "12/24/2016 20:18:58", "content": "The explanation from William Cukierski in the other thread suggests that there are two different types of document_ids, which seems unnecessarily vague to a naive observer as this could easily be clearer by simply labeling the two document_id types distinctly.
\n\nIf I understand this correctly there are two classes of document_ids: one referring to the document on which a display of ads is being presented to the user and one referring to the document to which the ad links when clicked. In other words, the flow of information is 1) user visits a document with document_id = A and this document has a display with various ads; 2) user clicks on an ad that has an associated document_id; 3) user is taken to document with document_id = B. If my understanding is correct then the document_id referred to in 2) would be document_id = B (the post-click document_id) and not document_id = A (the pre-click document_id).
\n\nIs this interpretation correct? If so, does this mean that the documents_meta, _topics, _entities, and _categories describe only one class of document_ids or both classes? It would make much more sense as to why the two classes were labeled the same if these descriptive tables house information describing both classes of documents.
", "rawMarkdown": "The explanation from William Cukierski in the other thread suggests that there are two different types of document_ids, which seems unnecessarily vague to a naive observer as this could easily be clearer by simply labeling the two document_id types distinctly.\r\n\r\nIf I understand this correctly there are two classes of document_ids: one referring to the document on which a display of ads is being presented to the user and one referring to the document to which the ad links when clicked. In other words, the flow of information is 1) user visits a document with document_id = A and this document has a display with various ads; 2) user clicks on an ad that has an associated document_id; 3) user is taken to document with document_id = B. If my understanding is correct then the document_id referred to in 2) would be document_id = B (the post-click document_id) and not document_id = A (the pre-click document_id).\r\n\r\nIs this interpretation correct? If so, does this mean that the documents_meta, _topics, _entities, and _categories describe only one class of document_ids or both classes? It would make much more sense as to why the two classes were labeled the same if these descriptive tables house information describing both classes of documents.", "votes": null } ], "comments": [ { "id": 138523, "author_name": "santyago", "author_url": "", "post_date": "10/09/2016 10:04:29", "content": "https://www.kaggle.com/c/outbrain-click-prediction/forums/t/24154/redundant-field-in-database
\n\n\n\n\nThe document_id associated with each ad_id in the promoted content is the document to which the ad refers (i.e. you would use this table in order to find the topics, categories, or entities referred to in an ad).
\n
I was also a little bit confused with promoted_content.document_id
\n\nGood luck!
", "votes": null, "replies": [] }, { "id": 138800, "author_name": "ceeker", "author_url": "", "post_date": "10/10/2016 22:31:30", "content": "\n\n\nIt's somewhat ambiguous to me what a document refers to.
\n
Document = Web Page. Here is the hierarchy:
\n\nHope this is clear
", "votes": null, "replies": [] }, { "id": 152292, "author_name": "rutledge", "author_url": "", "post_date": "12/24/2016 20:18:58", "content": "The explanation from William Cukierski in the other thread suggests that there are two different types of document_ids, which seems unnecessarily vague to a naive observer as this could easily be clearer by simply labeling the two document_id types distinctly.
\n\nIf I understand this correctly there are two classes of document_ids: one referring to the document on which a display of ads is being presented to the user and one referring to the document to which the ad links when clicked. In other words, the flow of information is 1) user visits a document with document_id = A and this document has a display with various ads; 2) user clicks on an ad that has an associated document_id; 3) user is taken to document with document_id = B. If my understanding is correct then the document_id referred to in 2) would be document_id = B (the post-click document_id) and not document_id = A (the pre-click document_id).
\n\nIs this interpretation correct? If so, does this mean that the documents_meta, _topics, _entities, and _categories describe only one class of document_ids or both classes? It would make much more sense as to why the two classes were labeled the same if these descriptive tables house information describing both classes of documents.
", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "138515": "promoted_content.csv provides additional details about the ad. There is also the \"document_id\" parameter. Does that document_id refer to details about the ad itself (describes the ad) or does it refer to a document where the ad is shown? \r\n\r\nIt's somewhat ambiguous to me what a document refers to. It definitely refers to what a user is visiting from the location bar in the browser, but it's not quite clear whether the same document set also refers to the 'ads' (which are essentialy referral documents, but paid for, so not ads in the typical sense), or whether it describes the documents to which the ads were assigned.\r\n\r\nCould someone clarify this please?", "138523": "https://www.kaggle.com/c/outbrain-click-prediction/forums/t/24154/redundant-field-in-database\r\n\r\n> The document_id associated with each ad_id in the promoted content is the document to which the ad refers (i.e. you would use this table in order to find the topics, categories, or entities referred to in an ad).\r\n\r\nI was also a little bit confused with promoted_content.document_id\r\n\r\nGood luck!", "138800": "> It's somewhat ambiguous to me what a document refers to.\r\n\r\nDocument = Web Page. Here is the hierarchy:\r\n\r\n1. A document is the largest container \r\n2. Within a document, there is a main content \r\n3. The document will host a display A display is a set of ads - this is randomized every time you visit that same page \r\n4. Each ad may or may not come from the same campaign, or advertiser \r\n5. There may be only 4 ads limited per display - could be more, but I have actually seen only 4 at a time\r\n6. A user visiting a document will generate a page view log\r\n7. A user clicking on a ad will generate a click log\r\n8. Every time the user visits a page, an event is generated and logged\r\n\r\nHope this is clear", "152292": "The explanation from William Cukierski in the other thread suggests that there are two different types of document_ids, which seems unnecessarily vague to a naive observer as this could easily be clearer by simply labeling the two document_id types distinctly.\r\n\r\nIf I understand this correctly there are two classes of document_ids: one referring to the document on which a display of ads is being presented to the user and one referring to the document to which the ad links when clicked. In other words, the flow of information is 1) user visits a document with document_id = A and this document has a display with various ads; 2) user clicks on an ad that has an associated document_id; 3) user is taken to document with document_id = B. If my understanding is correct then the document_id referred to in 2) would be document_id = B (the post-click document_id) and not document_id = A (the pre-click document_id).\r\n\r\nIs this interpretation correct? If so, does this mean that the documents_meta, _topics, _entities, and _categories describe only one class of document_ids or both classes? It would make much more sense as to why the two classes were labeled the same if these descriptive tables house information describing both classes of documents." }, "source": "meta" }