Results 1 to 7 of 7
  1. #1
    Join Date
    Aug 2003
    Location
    RI/ME
    Posts
    9

    Unanswered: new semantic data structure quandry

    Hi all:
    I'm in a bit of a quandry. My small company has recently developed a new semantic data structure based on simulation of dual reciprocating human memory structure and unified cognitive functionalities. We claim that human logic is structure determined and believe we have developed that, along with the unified theory of cognition algorithm.
    So? Here's my small problem: I'm the investor (big problem?) and I just received the attached note from a Chinese programmer we have working with us. It is difficult for me to understand, both in English language terms and in content terms. I'm not very fluent in data sorting principles nor in storage and memory space allocation technique.
    I've attached a very short note from him and was hoping to get a few people who are experts in this domain to check it out and give me an opinion as to what is the content, its meaning and significance. I'm hoping that we have something fiinancially significant and have been told with embedded logic, we are eventually able to displace or supplant a lot of traditional database design tasks, dba effort, that we can supply W3C with next gen semantic web architecture and perform natural language translation, to name a few...!!
    Thanks in advance,
    Wick Beavers
    Attached Files Attached Files

  2. #2
    Join Date
    Sep 2002
    Location
    UK
    Posts
    5,171
    Provided Answers: 1

    Re: new semantic data structure quandry

    Originally posted by Wick
    Hi all:
    I'm in a bit of a quandry. My small company has recently developed a new semantic data structure based on simulation of dual reciprocating human memory structure and unified cognitive functionalities. We claim that human logic is structure determined and believe we have developed that, along with the unified theory of cognition algorithm.
    So? Here's my small problem: I'm the investor (big problem?) and I just received the attached note from a Chinese programmer we have working with us. It is difficult for me to understand, both in English language terms and in content terms. I'm not very fluent in data sorting principles nor in storage and memory space allocation technique.
    I've attached a very short note from him and was hoping to get a few people who are experts in this domain to check it out and give me an opinion as to what is the content, its meaning and significance. I'm hoping that we have something fiinancially significant and have been told with embedded logic, we are eventually able to displace or supplant a lot of traditional database design tasks, dba effort, that we can supply W3C with next gen semantic web architecture and perform natural language translation, to name a few...!!
    Thanks in advance,
    Wick Beavers
    I don't understand all of the note, but the gist of it seems to be that instead of storing a street address as "700 Barberry Drive" you split it into 3 parts stored separately, with the "street" ('Barberry') and "suffix" ('Drive') held as foreign key IDs ("machine addresses") to lookup tables.

    If that is all it amounts to, you can do that with a standard database if you choose to:

    insert into street( street_id, street_name ) values (123, 'Barberry' );
    insert into suffix( suffix_id, suffix_name ) values (456, 'Drive' );
    insert into address( building_no, street_id, suffix_id( 700, 123, 456 );

    When given a new address such as '500 California Drive' you would first split it into the 3 parts 500, 'California' and 'Drive'. Then lookup the street record for 'California'; if not found, create one. Ditto the suffix.
    Finally, insert the address record using the street_id and suffix_id found (or created).

    This development may be a "new way to save data" in terms of a physical mechanism, but it is not true to imply that in a traditional database you are forced to store addresses as free text.

  3. #3
    Join Date
    Aug 2003
    Location
    RI/ME
    Posts
    9

    Store equivalent data once/ no data sorting

    In our semantic data structure, we store the word "Barberry" (or "Main" or "5th" or "Picadilly") only one time.
    Same goes for "Street", "Drive", "Circle"...
    If this is unique, one would save a lot of memory allocation space (that may explain the engineer's numbers on memory savings).
    If links are not subject to n squared problem (resolved, in part, by structure determined logic and application of Venn set theory, which provides specificity), I believe we overcome data sorting problem.
    Does this help further explain the attachment?
    Cheers and thanks for your time and thoughts.
    Wick

  4. #4
    Join Date
    Sep 2002
    Location
    UK
    Posts
    5,171
    Provided Answers: 1

    Re: Store equivalent data once/ no data sorting

    Originally posted by Wick
    In our semantic data structure, we store the word "Barberry" (or "Main" or "5th" or "Picadilly") only one time.
    Same goes for "Street", "Drive", "Circle"...
    If this is unique, one would save a lot of memory allocation space (that may explain the engineer's numbers on memory savings).
    Yes, and my point is that you are able to do that in a relational database also:

    select * from streets;
    ID Name
    123 Barberry
    124 California
    125 Forest
    126 High
    ...

    select * from suffixes;
    ID Name
    456 Drive
    457 Road
    458 Street
    ...

    select * from addresses;
    Number StreetID SuffixID
    700 123 456
    701 123 456
    120 124 456
    2 123 458

    Thus a typical addresses record consists of a small number of bytes, with the text description available via lookup into the other tables.

    Originally posted by Wick
    If links are not subject to n squared problem (resolved, in part, by structure determined logic and application of Venn set theory, which provides specificity), I believe we overcome data sorting problem.
    Does this help further explain the attachment?
    Not to me, no! I don't understand most of it. Sounds like you have a plan for breaking up an address into constituent parts and storing them in a physically ordered manner. Perhaps similar could be achieved in Oracle using index-organized tables based on my examples above?

  5. #5
    Join Date
    Aug 2003
    Location
    Delft, The Netherlands (EU)
    Posts
    447

    Re: Store equivalent data once/ no data sorting

    Hi Wick,

    You guys seemed to have re-invented the relational concept of normalization, but giving it difficult names. Please correct me, if I'm wrong, and explain, what's making your approach different.

    By the way, I just re-called a great story about three wizzkids, making a genious RDBMS ignoring conventional implementations of relational databases. See Alterian, and dive into their technology. You will like it!
    Last edited by DoktorBlue; 09-02-03 at 16:05.
    Make everything as simple as possible, but not simpler! - A. Einstein
    DB Problems? DB Explorer, BTrieve Re-engineering, DB Conversions & ETL? Conversion Tool

  6. #6
    Join Date
    Aug 2003
    Location
    RI/ME
    Posts
    9

    Struggle with me! Normalization versus dynamically defined relationships

    Hi:
    The way our semantic system works is inherently different than the traditional db/rdb. While dba's must define relationships and create rows, columns, tables and fields (static), ours does not require the implementation of these things to create dynamic relationships- we call the semantic data structure, "dynamic" for this reason, amonst others.
    From readers' opinions and comments, I am seeing that people are trying to "normalize" our semantic structure into a rd model- which is really not possible. I am speaking language of: "semantic data structure" and we are not communicating. All Roads Lead to Rome, so if I can state or demonstrate we arrive at the same place, then perhaps, I'm achieving part of my goal for this post.
    But I want to stress the structure I am discussing is as -or more- useful in the same application space as the static traditonal one you all use everyday. I need to prove that!
    This system is built on a new model of human cognition and simulated dual memory. A two column system only. When two column memory is unified with underlying cognitive functionalities, innate logic occurrs- just as in human brain. The two column memory contains a perceptual (data) memory and a conceptual (metadata) memory- conscious and subconscious, if you prefer. Multiple relations (one-to-many, many-to-one and many-to-many) can be established between the elements of two memories and when new data is introduced to existing ones ("is learned"), it falls automatically ing into dynamic relationships with existing data in memory. A Porsche is European, red, flashy, fast, high speed, two seater... the new word Porsche is categorized right along with existing data that has these same relationships and is categorized in the same group as Ferrari, Lamborghini, BMW and so on. It is NOT categorized with Mustang, GTO, Accura, Toyota or Corvette because it is not related with "European". You KNOW the difference! Thus, we also achieve auto-categorization and specificity in retrieval. What is "innate human logic" in terms of semantic data structure? It is the same functionality which allows us to establish logical dynamic relationships amongst the data we learn everyday. Also to retrieve knowledge with 100% accuracy. Further, it provides contextual disambiguation, and provides universal grammar.
    What I am trying to understand is if what I am claiming about the semantic data structure makes any sense to any db experts. I guess I'm talking a different language... Would love to find a common road to get to Rome.
    I'd be happy to send more information by email to anyone that might have a deeper interest. I've attached another document discussing how we may utilize this structure as a next gen semantic web architecture.
    Thanks again!
    Wick
    Attached Files Attached Files

  7. #7
    Join Date
    Sep 2002
    Location
    UK
    Posts
    5,171
    Provided Answers: 1

    Re: Struggle with me! Normalization versus dynamically defined relationships

    Originally posted by Wick
    Hi:
    The way our semantic system works is inherently different than the traditional db/rdb. While dba's must define relationships and create rows, columns, tables and fields (static), ours does not require the implementation of these things to create dynamic relationships- we call the semantic data structure, "dynamic" for this reason, amonst others.
    From readers' opinions and comments, I am seeing that people are trying to "normalize" our semantic structure into a rd model- which is really not possible. I am speaking language of: "semantic data structure" and we are not communicating. All Roads Lead to Rome, so if I can state or demonstrate we arrive at the same place, then perhaps, I'm achieving part of my goal for this post.
    But I want to stress the structure I am discussing is as -or more- useful in the same application space as the static traditonal one you all use everyday. I need to prove that!
    This system is built on a new model of human cognition and simulated dual memory. A two column system only. When two column memory is unified with underlying cognitive functionalities, innate logic occurrs- just as in human brain. The two column memory contains a perceptual (data) memory and a conceptual (metadata) memory- conscious and subconscious, if you prefer. Multiple relations (one-to-many, many-to-one and many-to-many) can be established between the elements of two memories and when new data is introduced to existing ones ("is learned"), it falls automatically ing into dynamic relationships with existing data in memory. A Porsche is European, red, flashy, fast, high speed, two seater... the new word Porsche is categorized right along with existing data that has these same relationships and is categorized in the same group as Ferrari, Lamborghini, BMW and so on. It is NOT categorized with Mustang, GTO, Accura, Toyota or Corvette because it is not related with "European". You KNOW the difference! Thus, we also achieve auto-categorization and specificity in retrieval. What is "innate human logic" in terms of semantic data structure? It is the same functionality which allows us to establish logical dynamic relationships amongst the data we learn everyday. Also to retrieve knowledge with 100% accuracy. Further, it provides contextual disambiguation, and provides universal grammar.
    What I am trying to understand is if what I am claiming about the semantic data structure makes any sense to any db experts. I guess I'm talking a different language... Would love to find a common road to get to Rome.
    I'd be happy to send more information by email to anyone that might have a deeper interest. I've attached another document discussing how we may utilize this structure as a next gen semantic web architecture.
    Thanks again!
    Wick
    Sounds a bit like "The Associative Model of Data" by Lazy Software (www.lazysoft.com), essentially a 2-table database of Things and Relationships between Things (read: a lot of nonsense).

    Also, someone called "Neo" I think has proposed something that sounds similar on comp.databases.theory (BTW, why are you proposing this in an Oracle forum rather than somewhere more general?)

    To me, this sounds like an interesting (if not novel) AI research idea, whereas generally business database designs are deliberately "static" so that meaning is imposed on the data, rather than letting information pile up and be interpreted by some kind of fuzzy logic.

    I may be wrong, it's outside my sphere of knowledge for sure.

    You'd probably get some lively and informed debate if you posted in comp.databases.theory rather than here.

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •