6

 [6]
After deduplication, the received CFi signals are sorted according to data type. As indicated above, CFi signals are typically delivered to the head end of the system core (e.g., cloud 410) with time, location and data type stamps attached to the payload data. One payload may represent simple text content (e.g., ASCII encoded) while another payload may represent simple sound content (e.g., .wav encoded) and yet another payload may represent bit-mapped encoded imagery (e.g., .bmp encoded). These different data types are sorted according to their data types so that sounds get stored adjacent to other sounds of the same general time-stamped period and/or of the same general location-stamped place and so that odor (smell) indicating signals get stored adjacent to other odor (smell) indicating signals of same place/time and so on. This is a first step in categorizing and parsing the possibly multi-typed ones of the received CFi signals. The goal is to form clusters of reasonably combinable CFi primitives that pass so-called, sanity checks before being used to build more complex combinations or clusterings of CFi signals. More specifically, if a musical-tone detecting sensor (not shown) at the user end (301A?) sends a first CFi packet holding 3 notes and then sends a second CFi packet holding 5 more notes, it is possible and likely that the total of 8 notes belong together as part of one melody; or perhaps they don't. Perhaps the latter 5 notes need to instead be clustered with the payload of yet a third, not yet, but to-be-sent CFi packet containing 7 further notes. In other words, there are a number of possible first level “permutations” here for clustering together received sequences of CFi signals, namely: (1) CFiPacket#1 (first 3 notes) belongs or does not belong as a prefix to CFiPacket#2 (next 5 notes); (2) CFiPacket#2 (the 5 notes) belongs or does not belong as a prefix to CFiPacket#3 (next 7 notes); (3) all of CFiPacket#1, #2 and #3 belong together as a continuous melody; (4) none of CFiPacket#1, #2 and #3 belong together as a continuous melody. The concept of forming likely “permutations” or clusters of alike CFi data signals; and then clusters of clusters will be explored in more detail later below.
First, and getting back to basics, it is to be understood that each of the CFi generating units 302 b? and 298 a? of FIG. 3D, as well as the local physical context reporting unit(s) 304/306, includes a current focus-indicator(s)/current context indicator(s) packaging subunit (not shown) which packages raw telemetry signals from the corresponding tracking sensors as typed data payloads into time-stamped, location-stamped, type-stamped, user-ID stamped, machine-ID stamped, and/or otherwise stamped and transmission ready data packets. These data packets are received by appropriate CFi-processing and context-indication processing servers in the head end (e.g., cloud) of the system core and processed in accordance with their user-ID (and/or local device-ID) and time and location and data type (and/or other stampings). In one embodiment, the CFi/context reporting signals sent to the head end are pre-packaged or re-packaged further downstream, after being transmitted, into hybridized signals, or so-called, HyCFi signals where additional context information beyond time, location and type is attached to the current focus indicating information, such as for example, identifications of other users in interactive proximity with the first user, where the latter can be indicative of a current social context in which the first user (301A?) finds him/herself to be situated within.
One of the basic processings that the data packet receiving servers (or automated services) perform at a front or downstream receiving part of the head end is to group (e.g., cluster and/or cross-associate with logical links) the separately received packets of respective users and/or of data-originating devices according to user-ID (and/or according to local originating device-ID and/or data-type ID) and to also group received packets belonging to different times of origination and/or different times of transmission into respective chronologically ordered groups of alike types of data. In other words, musical note signals get grouped with other musical note signals, image defining signals get grouped with other and alike (e.g., .bmp, .jpg, .mp3) image defining signals and so on. The so pre-processed CFi signals are then normalized by normalizing modules like 302 qe?-302 qe 2? if the signals had not been yet normalized (e.g., de-idiosyncratized) earlier downstream. Then the normalized CFi and/or context indicating signals are fed into CFi clustering, cross-associating and categorizing-mechanisms 302? and 298? provided further upstream for yet further processing. (This further processing will be explained shortly but later below). At this stage it is understood that the muddled streams of data from different users and different ones of their local sensors have been untangled and purified, so to speak, such that the CFi data payloads of a first user, UsrA have been sorted out and stored in a storage area associated with user UsrA while the CFi data payloads of a second user, UsrB have been sorted out and stored in a storage area associated with that second user, UsrB. Moreover, for each user (for each persona of each user), the received CFi data payloads have further been chronologically and type wise and location wise been untangled and purified, so to speak, such that musical notes data picked up by a respective first musical-notes sensor are grouped together with one another in a correct time ordered manner and such that musical notes data picked up by a respective second musical-notes sensor (at a different location) are grouped together with one another in a correct time ordered manner, and the so-ordered data sets are further organized relative to one another in chronologically and type wise and location wise manner, and so on. More specifically, for the given example, the first and second musical-notes sensors may be differently placed microphones within an orchestra and the picked up notes may be from different musical instruments (e.g., piano, violin, clarinet) where the orchestra is playing harmonized stanzas which respectively are intended to be cognitively perceived in organized combinations or clusterings. Therefore one of the intended functions of a CFi's storing and organizing space such as 302? is to store in context appropriate organizations, CFi signals whose represented physical counterparts were intended by the user (301A?) or another to be cognitively perceived in relative unison.
The first set of sensors 298 a? have already been substantially described above (as eyeball movement trackers, head direction trackers, etc.). A second set of sensors 302 b? (referred to here as attentive-outputting tracking sensors) are also provided and appropriately disposed for tracking various expression outputting (code outputting) actions of the user, such as the user uttering in-context words (301 w), consciously nodding or shaking or wobbling his head, typing on a keyboard, making apparently-intentional hand gestures, clicking, tapping or otherwise activating different activateable data objects displayed on his screen and so on. As in the case of facial expressions that show attentive inputting of user accessible content (e.g., what is then displayed on the user's computer screen and/or played through his/her earphones even though the user may not watch it or listen to it), unique and abnormal output expressions (e.g., pet names for things, pre-coded combinations of tongue projections and other actions, a.k.a. hot-keying gestures) are run through expression-translating lookup tables (LUT's) and/or knowledge base rules (KBR's) of then active PEEP, CpCCp and/or other profiles for translating such raw expressions into more normalized (less idiosyncratic), Active Attention Evidencing Energy (AAEE) indicator signals of the outputting kind. In one embodiment, the in-context uttered words of the user are supplied to an automated speech recognition module (not shown) that automatically uses context (e.g., signal 316 o) in combination with speech pattern matching to then generate semantic codings representing the user uttered words in a textual and/or other more readily processible manner. The so-generates, semantic codings of the user's raw outputs form part of the “normalized” output signals of the user. The normalized AAEE indicator signals 298 e? of the inputting kind have already been described above. One example, by the way, of the normalization of abnormal output expressions may occur when the respective user is a multilingual user and is using an uncommon foreign language whereas keyword expressions then being received by the head end are pre-characterized as needing to belong to one agreed-upon standard language (e.g., English). In that case, words that the respective user may inadvertently output in a non-standard language are automatically translated into the agreed-upon standard language (e.g., English).
The normalized Active Attention Evidencing Energy (AAEE) signals, 302 e? and 298 e? are next inputted into corresponding first and second CFi clustering/categorizing mechanisms 302? and 298? as already mentioned. These clustering/categorizing mechanisms organizingly store the separately received CFi signals (302 e? and 298 e?) into yet more finely categorized and usable groupings (clusterings and/or categories) than just having them grouped according to user-ID and/or time or telemetry origination and/or location of telemetry origination. The further organizing of the received CFi signals (302 e? and 298 e?) is carried out with aid of so-called, CFi categorizing, clustering and inferencing engines 310? that connect in a feedback loop manner to the CFi clustering/categorizing spaces (mapping mechanisms) 302? and 298? and also in a feedback loop manner to other system-maintained mapping mechanisms (e.g., to content source space 314? (css), to context space (xs), to emotions space (es), and so on). One form of such finer categorizing of the received CFi signals (302 e? and 298 e?) is to parse them as being limbically-directed CFi's (example: “Please can't we just all get along without engaging in ad hominem attacks?”), as being neo-cortically directed CFi's (example: “Those numbers do not add up.”) or as being more primitive cognitions (example: “You have me blowing coffee out of my nostrils and laughing out loud (LOL)”). Another form of such finer categorizing of the received CFi signals (302 e? and 298 e?) is to parse them as being loosely directed to one broad topic domain or another (example: the liberal arts versus the math and science arts). Additionally, the finer categorizing of the received CFi signals (302 e? and 298 e?) includes parsing them according to more likely groupings (clusterings) and less likely combinatorial assemblages.
This latter part of the improved grouping/clustering process provided by the CFi categorizing, clustering and inferencing engines 310? is best explained with a few yet more specific examples. Assume that within the 302 e? signals (AAEE outputting signals) of the corresponding user 301A? there are found three keyword expressions: KWE1, KWE2 and KWE3 that have been input into a search engine input box, one at a time over the course of, say, 9 minutes. (The latter timings can be automatically determined from the time stamps of the corresponding CFi data packet signals that carry the keyword payloads.) One problem for the CFi categorizing mechanism 302? (and its clustering/organizing engines 310?) is how to resolve whether each of the three received and stored keyword expressions: KWE1, KWE2 and KWE3 is directed to a respective separate topic or whether all three are directed to a same topic such that they should be processed as the full combination of all three keywords or whether some other permutation holds true (e.g., KWE1 and KWE3 are directed to one topic but the time-wise interposed KWE2 is directed to an unrelated second topic—or is just a nonsense word inadvertently thrown in to the sequence of events). This is referred to here as the CFi grouping and parsing problem. Which CFi's belong with each of the others and which belong to another group or stand by themselves or do not belong at all (and thus deserve to be ignored)? By way of a more specific example, assume that KWE1=“Lincoln” and KWE3=“address” while KWE2=“Goldwater” although perhaps the user (a Fifth Grade student) intended a different second keyword such as “Gettysburg”. (Note: At the time of authoring of this example, a Google™ online search for the string, “lincoln goldwater address” produced zero matches while “lincoln gettysburg address” produced over 500,000 results. An educated human being can quickly see that the example of KWE2=“Goldwater” does not belong. It makes no sense. But for a computer, the problem may not be easily spotted and resolved.
A second problem for the CFi clustering/categorizing mechanism 302?/310? is how to resolve what kinds of CFi signals is it receiving in the first place? How did it know that expressions: KWE1, KWE2 and KWE3 were in the “keyword” category, as opposed to, for example, in the URL's category? In the case of keyword expressions, that question can be resolved fairly easily because the exemplary KWE1, KWE2 and KWE3 expressions are detected as having been submitted to a search engine through a search engine dialog box or a search engine input procedure. But other text-based CFi's and more to the point, non-textual CFi's, can be more difficult to categorize. Consider for example, a nod of the user's head up and down by the user and/or a simultaneous grunting noise made by the user. What kind of intentional expression, if at all, is that? The answer depends at least partly on context, culture and/or user mood. If the most recent context state of the user is determined by the STAN—3 system 410 (by output signal 316 o in FIG. 3D) to be one where the user 310A? is engaged in a live video web conference with other persons of a Western culture, then the up-and-down head nod may be taken as an expression of intentional affirmation (yes, agreed to) being communicated to the others if the nod is pronounced enough. On the other hand, if the user 301A? is simply reading some text to himself (a different social context, namely, being alone) and he nods his head up and down or side to side and with less pronouncement, that may mean something different, dependent on the currently active PEEP profile of the respective user. The same would apply to the grunting noises or other non-textual user outputs.
In general, the CFi receiving and clustering/categorizing mechanisms 302?/298? and the interconnected engines 310? first cooperatively assign incoming CFi signals (e.g., normalized/augmented CFi signals) to one or the other or both of two mapping mechanism parts, the first being dedicated to handling information outputting activities (302?) of the user 301A? and the second being dedicated to handling more passive information inputting activities (298?) of the user 301A?. If the CFi receiving and categorizing mechanisms 302?/298?/310? cannot parse as between the two, they copy the same received CFi signals to both sides. Next, the CFi receiving and categorizing mechanisms/engines 302?/298?/310? try to categorize the received CFi signals into predetermined subcategories unique to that side of the combined categorizing mapping mechanism 302?/298?. Keywords versus URL expressions would be one example of such categorizing operations. In this case, both of keywords and URL's belong to a broader class of sequential textual content (which could include sequentially supplied codes or symbols as well as traditional alphanumeric characters). Musical notes versus random background noise may be another example of CFi's of different categories. (Ultimately, musical background notes might be mapped as corresponding to communally-created and communally-accepted music primitives having data structures such as shown in FIG. 3F. However, the present discussion is not yet ripe enough to deal with that eventuality. It will be taken up later below.)
URL string expressions can be automatically categorizing as such (as being Universal Resource Locator type expressions) by their prefix and/or suffix and/or in-fix strings (e.g., by detection of having a “dot.com” character string embedded therein or having the “at mark” symbol infixed therein if it is an email address for example). Other such categorization parsings include but are not limited to: distinguishing as between meta-tag type CFi's, image types, sounds, emphasized text runs (e.g., those that are italicized, bolded, underlined, etc.), body part gestures, topic names, context names (i.e. role undertaken by the user), physical location identifications, platform identifications, social entity identifications, social group identifications, neo-cortically directed expressions (e.g., “Let X be a first algebraic variable . . . ”), limbically-directed expressions (e.g., “Please, can't we all just get along?”), and so on. More specifically, in a social dynamics subregion of a hybrid topic and context space, there will typically be a node disposed hierarchically under limbic-type expression strings and it will define a string having the word “Please” in it as well as a group-inclusive expression such as “we all” as being very probably directed to a social harmony proposition. In one embodiment, expressions output by a user (consciously or subconsciously are automatically categorized as belonging to none, or at least one of the following layers of a triune brain model: (1) neo-cortically directed expressions (i.e., those appealing to the intellect), (2) limbically-directed expressions (i.e., those appealing to social interrelation attributes) and (3) reptilian core-directed expressions (i.e., those pertaining to raw animal urges such as hunger, fight/flight, etc.). In one embodiment, the neo-cortically directed expressions are automatically allocated for processing at least by the topic space mapping mechanism 313? because expressions appealing to the intellect are generally categorizable under different specific topic nodes. In one embodiment, the limbically-directed expressions are automatically allocated for processing at least by the emotional/behavioral states mapping mechanism 315? because expressions appealing to social interrelation attributes are generally categorizable under different specific emotion and/or social behavioral state nodes. In one embodiment, the reptilian core-directed expressions are automatically allocated for processing by at least a biological/medical/emotional state(s) mapping mechanism (315?, see also exemplary primitive data object of FIG. 3O) because raw animal urges are generally attributable biological states (e.g., fear, anxiety, hunger, etc.). More will be said about parsing of CFi's into clusters and clusters of clusters when the discussion reaches FIG. 3U. The above is more in the way of an introduction.
As mentioned, the automated and augmenting categorization of incoming CFi's is performed with the aid of one or more CFi clustering/categorizing and inferencing engines 310? where the inferencing engines 310? have access to categorizing nodes and/or subregions within, for example, to parts within topic and/or context space and/or within biological states space (e.g., in the case of the social harmony invoking example given immediately above: “Please, can't we all just get along?”) or more generally, access to categorizing nodes and/or subregions within the various system-maintained Cognitive Attention Receiving Spaces (CARSs). The inferencing engines 310? receive as their inputs, last known state signals (e.g., 316 o) from various ones of the state mapping mechanisms (CARSs) as representing rough indications of associated CARSs points cross-correlating to current CFi clusters and indirectly, the respective user's state of mind. More specifically, the last determined to be most-likely context states are represented by “xs” signals received by the inferencing engines 310? from the output 316 o of the context mapping mechanism 316?; the last determined to be most-likely focused-upon sub-portions of content materials are represented by “css” signals received from the output 3140 of the content source space mapping mechanism 314? (where 314? stores pointers to (e.g., URL's to), or abbreviated representations of content that is likely available to be currently focused-upon by the user 301A?); the previously determined to be most-likely CFi clusterings/categorizations are received as currently stored “HyCFis” signals from the CFi categorizing mechanism 302?/298?; the last determined as probable emotional/behavioral states of the user 301A? are received as “es” signals (emo signals) from an output 315 o of an emotional/behavioral state mapping mechanism 315?, and so on.
In one embodiment, the inferencing engines 310? operate on a weighted assumption that the past is a good predictor of the present and of the near future. In other words, the most recently determined states “xs”, “es”, “HyCFi's of the respective CFi's from the one user (or of another social entity that is being processed) are first used for categorizing the more likely categories for next incoming new CFi signals 302 e? and 298 e?. The “css” signals tell the inferencing engines 310? what content was logically available (e.g., on a nearby TV screen—by looking up TV show scheduling databases, on a nearby computer screen, via nearby loudspeakers or earphones, etc.) to the user 310A? at the time one of the CFi's was generated (time and place stamped CFi signals—see 30U.10 of FIG. 3U) in regard to content then being presented for potential perception by the respective user. More specifically, if a search engine input box was displayed in a given screen area, and the user inputted a character string expression into that area at that time, then the expression is determined to most likely be a keyword-based search expression (KWE). If a particular sound was being then outputted by a sound outputting device near or on the user, then a detected sound at that time (e.g., music) is determined to most likely be a music and/or other sound CFi the user was exposed to at the time of telemetry origination. By categorizing the received (and optionally normalized/de-idiosyncraticized) CFi's in this manner it becomes easier to subsequently group likes with alikes and parse them, and cluster logically interrelated ones of them together so as to build clusters of them (or clusters of clusters) before transmitting the parsed and grouped/clustered (and optionally hybridized) CFi's as input vector signals (e.g., HyCFi's) into appropriate ones of the mapping mechanisms (e.g., 313?, 316?) for further processing.
Yet more specifically and by way of example, it will be seen below that the present disclosure contemplates a music-objects organizing space (or more simply a music space, see FIG. 3F). Current background music that is available to the user 301A? may be indicative of current user context and/or current user emotional/behavioral state (e.g., mood). Various nodes and/or subregions in music space can logically link to ‘expected’ emotional/behavioral state nodes, and/or to ‘expected’ context state nodes/regions and/or to ‘expected’ topic space nodes/regions within corresponding data-objects organizing spaces (mapping mechanisms). An intricate web of cross-associations is quickly developed simply by detecting, for example, a musical melody being played in the background, determining that it is a musical melody, and inferring from that determination, a host of parallel one of more likely possibilities. More to the point, if the user 301A? is detected as currently being exposed to soft calming music, the ‘expected’ emotional/behavioral state of the user is automatically assumed by the CFi categorizing and inferencing engines 310? (in one embodiment and with use of the music space (not shown in FIG. 3D) and its cross-associating links to emotional/behavioral state space 315?) to be a calm and quieting one. That colors how other CFi's received during substantially the same time period and in substantially the same physical context (XP) will be categorized because the user's mood generally determines the currently activated PEEP record (part of 301 p?) for that user. Each CFi categorization can assist in the additional and more refined categorizing and placing of others of the contemporaneous and/or co-located CFi's of a same user in proper context since the other CFi's were received from a same user and in close chronological and/or geographical interrelation to one another where user non-physical context (more cerebral context) is safely assumed to be a steady state one.
Aside from categorizing individual ones of the incoming CFi's as being one type or another (e.g., textual versus melodic), the CFi clustering/categorizing and inferencing engines 310? parse and group (cluster) the incoming CFi's as either probably belonging together with each other or probably not belonging together. It is desirable to correctly group together emotion-indicating CFi's with their cross-associated non-emotional CFi's (e.g., keywords, URL's) because that is later often used by the system to determine how much “heat” a user is casting on one node or another in topic space (TS) and/or in other such spaces (e.g., keyword space, URL space, and so on). More specifically, if biological state telemetry indicates the user's heart rate has suddenly increased, his/her respiration level has increased, and the user's current PEEP record indicates that this user tends to experience such increase of heart rate (e.g., beats per minute) approximately 10 seconds after having visually perceived emotionally-inciting content, the system can then logically cross-associate the later-in-time, fight-or-flight reaction (e.g., increased heart rate/increased respiration rate) with content that was presented to the same user 10 seconds ago. Consequently, that content, and/or the URL of the site from which it was presented, are given enhanced “heat” signatures.
In terms of a yet more specific example, consider again the sequentially received set of keyword expressions: KWE1, KWE2 and KWE3; where as one example, KWE1=“Lincoln”, KWE3=“address” while KWE2 is something else and its specific content may color what comes next. More specifically, consider how topic and context may be very different in a first case where KWE2=“Gettysburg” versus an alternate case where KWE2=“car dealership”. Those familiar with contemporary automobile manufacture would realize that “Lincoln car dealership” probably corresponds to a sales office of a car distributor who sells on behalf of the Mercury/Lincoln™ brand division of the Ford Motor Company. “Gettysburg Address” on the other hand, corresponds to a famous political event in American history. These are usually considered to be two entirely different topics and normally would have two separate nodes or subregions in topic space, although a topic node covering both at the same time is possible.
Assume also that about 90 seconds after KWE3 was entered into a search engine and results were revealed to the user, the user 301A? became “anxious” (as is evidenced by subsequently received physiological CFi's; perhaps because the user is in Fifth Grade and just realized his/her history teacher expects the student to memorize the entire “Gettysburg Address”). A question for the machine system to resolve in this example is which of the possible permutations of KWE1, KWE2 and KWE3 plus the emotion-indicating CFi that followed form a cross-associated cluster indicating there is a specific keyword expressions clustering (where the latter clustering in keyword space points to a corresponding topic in topic space—see keyword to topic link 370.6 of FIG. 3E) and indicating that the user became “anxious” over this keyword cluster/topic (or other subpart of another CARS), whereby the system should then record a projection of increased “heat” on the associated keyword nodes or cross-associated topic nodes (or nodes of other spaces)? Was it KWE1 taken alone or all of KWE1, KWE2 and KWE3 taken in combination or a subcombination of that? For sake of example, let it be assumed that KWE2 (e.g., =“Goldwater”) was a typographic error inputted by the user. He meant at the time to enter KWE3 instead, but through inadvertence, he caused an erroneous KWE2 to be submitted to his search engine. In other words, the middle keyword expression, KWE2 is just an unintended noise string that got accidentally thrown in between the relevant combination of just KWE1 and KWE3. How does the system automatically determine that KWE2 is an unintended noise string, while KWE1 and KWE3 belong together? The answer is that, at first, the machine system 410 does not know. However, embedded within a keyword expressions space (see briefly 370 of FIG. 3E) there will often be spatially “clustered” and combinatorial sets of keyword expressions (in layer 371 as shall be explained below) that are predetermined to likely make semantic sense (e.g., where the keyword combination might be represented by “operator” node 373.1 of FIG. 3E) and missing from that space will be nodes and/or subregions representing combinatorial sets of keyword expressions (e.g., “KWE1, AND KWE2 AND KWE3”) that are not predetermined to make semantic sense (at the relevant time; because after this disclosure is published, the phrase, “lincoln goldwater address” might become attributable to a corresponding topic of a STAN system). Incidentally, it is to be understood that the keyword expressions data-objects organizing space (370) is merely an example of other data-objects organizing spaces including data-objects storing spaces whose stored signals represent other textual expression strings (e.g., URL's, meta-tags, etc.) besides just spatially clustered keyword expression strings. This will be further detailed when the textual string primitive 30W.0 of FIG. 3W is explained later below. As mentioned above, “primitives” are data structures that can be used and combined to build more complex data structures by means of operator nodes where the more complex data structures represent more complex cognitions while the “primitives” represent relatively simple cognitions of one form (e.g., linguistic) or another (e.g., visual, melodic, etc.).
It should be recalled at this juncture that the inferencing engines 310? of FIG. 3D have access to the hierarchical data structures stored inside various ones of the system's data-objects organizing spaces (mapping mechanisms, a.k.a. Cognitive Attention Receiving Spaces). Accordingly, the inferencing engines 310? can first automatically and on a trial and error basis, entertain the possibility that the keyword permutation: say, “KWE1, AND KWE2 AND KWE3” can make semantic sense to a reasonable or rational STAN user situated in a context similar to the one that the CFi-strings-originating user, 301A? is situated in. Accordingly, the inferencing engines 310? are configured to automatically search through a hybrid context-and-keywords space (not shown, but see briefly in its stead, node 384.1 of FIG. 3E) for a pre-existing node corresponding to (matching to, or strongly cross-correlating to, namely, being substantially same or similar to it—which concept of substantially similarity will be explained elsewhere herein—) the entertained permutation of the combined CFi's and it then discovers that the in-context node corresponding to the entertained first permutation (a first trial balloon, see also 30V.12 of FIG. 3V): “KWE1, AND KWE2 AND KWE3” is not there (or has a very low approval rating by the mainstream of users—it does not meet with strong communal consensus as being a reasonable combination). As a consequence, the inferencing engines 310? may automatically throw away the entertained first permutation (e.g., “Lincoln's Goldwater Address”) as being an unreasonable/irrational one (unreasonable or lacking sanity at least to the machine system at that time) or the system will shuffle it to a bottom of a list of more likely permutations for reconsideration at a later time; and if the machine system is properly modeling a reasonable/rational person of a relevant system sub-community where that modeled person is similarly situated in a context close to that of user 301A?, the rejected/downgraded keyword permutation will also be deemed unreasonable to the similarly situated reasonable person. In one embodiment, the so-called, sanity check for trial permutations (e.g., trial clusterings of keywords) includes an automated test for cross-correlation as between textual or phonetic content and nodes of a system-maintained linguistic space (see FIG. 3I). More specifically, the close mixing of an adverb and adjective (e.g., the “quickly brown fox”) might indicate that something is not quite right with a trial permutation because a noun should not be normally modified by an adverb, although the present disclosure is open to the idea that new forms of cognition may arise with time wherein such rules might be properly violated once such violation is accepted by the relevant community.
In one embodiment, the inferencing engines 310? alternatively or additionally have access to one or more online search engines (e.g., Google™ Bing™) and/or Wiki-sites (e.g., Wikipedia™) and the inferencing engines 310? are configured to submit some of their entertained keyword permutations to the one or more online search engines and/or wiki engines (and in one embodiment, in a spread spectrum fashion so as to protect the user's privacy expectations by not dishing out all permutations of all CFi clusters to just one search/wiki engine) and to determine the quality (and/or quantity) of matches found so as to thereby perform a sanity check and automatically determine the likelihood that the entertained keyword permutation is a relatively valid one (e.g., one that can make semantic sense) as opposed to being a set of unrelated terms which combination is not worthy of prioritized consideration at the moment. However, in discovering that one permutation of, say plural keywords has more search engine hits than another, the inferencing engines automatically discount the popularity of shorter keyword permutations versus longer ones (ones with more terms to match) because, of course; the shorter ones are more likely to have a larger number of hits. For example, the one keyword, “Lincoln” will typically draw a much larger number of hits (matches) than the more defined, two word permutation of “Lincoln AND Address”. In one embodiment, the system is configured to prefer medium sized clusters of roughly three words each (or more specifically, in the range of two words minimum and five words maximum as an example); e.g., “Lincoln AND Gettysburg AND Address” over one word clusters and over say, 7 word clusters. The reason is because it has been found that the human brain works best in building up concepts as singlets, doublets and triads of linguistic cognitions (e.g., “the”/“quick brown fox”/“jumped over”).
More generally speaking, the inferencing engines 310? function as trial permutation generating engines which generate different trial permutations of clustered or otherwise grouped together CFi's or HyCFi's and then test the generated permutations for cross-correlation strengths relative to search engine results for the same trial permutations and/or for cross-correlation strengths relative to best-matched points, nodes or subregions of system-maintained/stored Cognitive Attention Receiving Spaces (CARSs), where respective cross-correlation strength scores are then assigned to the tested CFi and/or HyCFi permutations (and discounted for the unfair advantage that short permutations have over longer ones). The scored permutations are then sorted and stored as a sorted list. A subset of the scored permutations that have comparatively highest scores (after discounting for length and number of words) are then used to identify corresponding ones of the CARSs and points, nodes or subregions within them as being most likely ones of such portions of the system-maintained CARSs to which the received and test-wise clustered CFi's belong (see briefly, cluster definer 30U.12 in FIG. 3U). These results are represented in FIG. 3D by output signals 311? of the inferencing engines 310?. The corresponding, and once-clustered CFi's (the highest scoring permutations, including clusters of clusters) are then applied as search inputs into the identified portions of the system-maintained CARSs, often together with the current context-indicating signals 316 o so that context-relevant results (e.g., invitations to chat rooms) will next be developed and so that, optionally, clusters of clusters of the CFi's (see briefly, cluster definer 30U.14 in FIG. 3U) can next be developed with use of enlightening results produced by the first round of mappings into the various Cognitions-representing Spaces.
In terms of a more specific example, if the permutation of “Lincoln's Address” (“KWE1 AND KWE3” of the above example where KWE2 is ignored) receives the highest, post-discount cross-correlation scores, that permutation is combined with demographic context information indicating, for example, that the respective user is a Fifth Grade student now trying to do his/her history homework. The context-augmented search permutation is then applied for example, as an input vector into the topic space mapping mechanism 313? with instructions to find the best matching nodes or subregions for that context-augmented search permutation (e.g., a Fifth Grade Student doing homework re a so-called, “Lincoln's Address”). Those will likely lead to topic nodes that are relevant to the specific user and his/her current areas of focus. It is within the contemplation of the present disclosure to repeat the above for creating sorted lists of hybrid-wise clusters of clusters (e.g., “KWE1 AND KWE3” AND “URL5 AND URL7”); and then clusters of clusters of clusters and so on.
Stated in other words, eventually, the inferencing engines 310? will have automatically built up and entertained a more complex keyword permutation represented for example by “KWE1 AND KWE3 AND Context=user's current context” (e.g., “Lincoln's Address for purposes of a Fifth Grade Student”) of the above given example. Then, according to this example, the inferencing engines 310? determine the probable sanity of this more complex keyword permutation by trying to find one or more corresponding nodes and/or subregions in keyword and context hybrid space (e.g., cross-correlating strongly with “Lincoln's Address”) and/or many search hits from the utilized online search engines (e.g., Google™, Bing™) where some nodes and/or hits are identified as being more likely than others to be applicable, given the demographic context of the user 301A? who is being then tracked (e.g., a Fifth Grade student). This tells the inferencing engines 310? that the “KWE1 AND KWE3” permutation is a reasonable one that should be further processed (ahead of other less likely, more lowly scored permutations) by the topic and/or other mapping mechanisms (313? or others) so as to produce a current state output signal (e.g., 313 o) corresponding to that reasonable-to-the-machine keyword permutation (e.g., “KWE1 AND KWE3”) and corresponding to the then applicable user context (e.g., a Fifth Grade student who just came home from school and normally does his/her homework at this time of day). One of the outcomes of determining that “KWE1 AND KWE3” is a more likely to be valid permutation while “KWE2 AND KWE3” is not or is an unlikely to be sensible one (because KWE2 is accidentally interjected noise) is that the timing of emotion development (e.g., user 301A? becoming “anxious”) can be cross-associated as likely to have begun either with the results obtained from user-supplied keyword, KWE1 or the results obtained from KWE3 but not from the time of interjection of the accidentally interjected KWE2. That outcome may then influence the degree of “heat” and the timing of “heat” cast on topic space nodes and/or subregions that are next logically linked to the keyword permutation of “KWE1 AND KWE3”. Thus it is seen how the CFi-permutations testing and inferencing engines 310? can help form reasonable groupings or clusterings of keywords and/or other CFi's that deserve prioritized further processing while filtering out unreasonable groupings that will likely waste processing bandwidth in the downstream mapping mechanisms (e.g., topic space 313?) without likely producing useful results (e.g., valid topic identifying signals 313 o).
The grouped (e.g., clustered or cross-associated and thus parsed) and categorized CFi permutations are then selected and applied for further testing against nodes and/or subregions in what are referred to here as either “pure” data-objects organizing spaces (e.g., like topic space 313?) or “hybrid” data-objects organizing spaces (e.g., 397 of FIG. 3E) where the nature of the latter will be better understood shortly. By way of at least a brief introductory example here (one that will be further explicated in conjunction with FIG. 3L), there may be a node in a music-context-topic hybrid space (see 30L.8 of FIG. 3L) that back links to certain subregions of topic space (see briefly 30L.8 c-e of FIG. 3L). (Example: What musical score did the band play just before Abraham Lincoln gave his famous “Gettysburg Address”?) If the current user's focal state (see briefly focus-identifying data object 30K.0? of FIG. 3L) points to the hybrid, in-context music-topic node, it can be automatically determined from that, that the machine system 410 should also link back to, and test out, the topic space region(s) of that hybrid node to see if multiple hints or clues (e.g., clusters of clusters of hybridized CFi's) simultaneously point to the same back-linked topic nodes and/or subregions. If they do, the likelihood increases that those same back-linked topic nodes and/or subregions are focused-upon regions of topic space corresponding to what the user 301A? is truly focused-upon and corresponding focus scores for those nodes/subregions are then automatically increased. At the end of the process, the added together plus or minus scores for different candidate nodes and/or subregions in topic space (or other space) are summed and the results are sorted to thereby produce a sorted list of more-likely-to-be focused-upon topic nodes (or subregions) and less likely ones. Thus, a current user's focus-upon a particular subregion of topic space can be determined by an automated machine means that operates with artificial intelligence (AI) types of software to arrive at context-appropriate determinations regarding what topics are more likely than not to be the areas of focus of the respective user. As mentioned above (with regard to output signal 313 o; most likely topics), the sorted results list will typically include or be logically linked to the user-ID and/or an identification of the local data processing device (e.g., smartphone) from which the corresponding CFi streamlet arose and/or to an identification of the time period in which the corresponding CFi streamlet (e.g., KWE1-KWE3) arose. (See also briefly, CFi data structure 30U.10 of FIG. 3U.) Hence, physical context for the CFi streamlet (e.g., KWE1-KWE3) is often present and the CFi permutations testing process often works with hybridized current focus indicators (HyCFi's) in which the attention giving activities/states of the user are cross-associated with physical context representing signals (XP, generated by module 304 for example) indicative at least of current physical context of the user. Accordingly, the input planes of CFi processing mechanisms 302? and 298? in FIG. 3D are illustrated with the parenthetical notation, “(+XP)” to indicate that, in general (there can be exceptions), received CFi signals (302 e? and 298 e?) are of the with-context-appended hybridized type of current focus indicators (HyCFi's) so that at least current physical context “(+XP)” is generally included in the consideration of which permutations of separately received CFi signals are most likely to belong together as a reasonably parsed clusterings or groupings of such received CFi signals and which are not.
Still referring to FIG. 3D, aside from the topic space mapping mechanism 313? and the context space mapping mechanism 316?, only a few others of the more frequently usable ones of many possible data-objects organizing (mapping) spaces (e.g., Cognitive Attention Receiving Space mapping mechanisms) are shown in FIG. 3D. These include the then-available-to-user-content space mapping mechanism 314?, the emotional/behavioral user state mapping mechanism 315?, and a social interactions theories mapping mechanism 312?, where the last inverted pyramid (312?) in FIG. 3D can be taken to represent yet more such spaces.
Referring yet a bit longer to FIG. 3D, it is to be understood that the automated matching of STAN users with corresponding chat or other forum participation opportunities and/or the automated matching of STAN users with suggested on-topic content (or other informational resources such as topic-knowledgeable other users/experts) is not limited to having to isolate specific nodes and/or subregions in just topic space 313?. STAN users can be automatically matched to one another and/or invited into same chat or other forum participation sessions on the basis of substantial commonality as between either their raw CFi signals (298 e?, 302 e?) or their normalized, clustered and/or categorized CFi's of a recent time period or the fact that their raw or normalized, clustered and/or categorized CFi's best fit with roughly same subregions in one or more of the system-maintained Cognitions-representing Spaces. In FIG. 3D, this possibility is represented by CFi's storing subregion CFiSR1 inside pyramid 302?. CFi's that cluster within this one region may attach to a so-called, CFi's Collecting Node (CFiSRO 30U.0 in FIG. 3U) where the node points to associated chat or other forum participation opportunities (see fields 30U.6, 30U.7) or associated other informational resources (30U.8). In other words, just the clusterings of CFi's can be used to refer a given STAN user to another given STAN user and/or to specific online content or other informational resources (for further research) due to the substantial matching between the raw or categorized CFi's of that user in a recent time period and correspondingly cross-matched nodes and/or subregions in spaces other than topic space, such as for example, in a keyword expressions space (not shown in FIG. 3D, see instead FIG. 3E). Alternatively or additionally, STAN users can be automatically matched to one another and/or invited into same chat or other forum participation sessions on the basis of substantial commonality as between nodes and/or subregions of other-than-topic space spaces that their raw or categorized CFi's point towards (cross-correlate to with relatively high cross-correlation scores based on context as well as other attributes). The CFi's of cross-introduced STAN users do not have to point to exactly the same topic node (as an example) in topic space for the users to be introduced to one another. Instead, the CFi's can merely point to points, nodes or subregions (PNOSs) in topic space (and/or in another such space) where the pointed to PNOS's are deemed substantially close to one another in a hierarchical and/or spatial sense based on predefined closeness rules stored for the corresponding subregion of the respective space. (In other words, close enough within that context.)
Stated in alternative words, topic space is not the one and only means by way of which STAN users can be automatically joined together based on the CFi's up or in-loaded on their behalf into the STAN—3 system core from their local monitoring devices. The raw CFi's alone (298 e?, 302 e?) or normalized ones may provide a sufficient basis by themselves for automatically generating invitations and/or suggesting additional content for the users to look at. It will be seen shortly in FIG. 3E that nodes in non-topic spaces (e.g., keyword expressions space) can logically link to topic nodes and that those non-topic nodes can of themselves similarly point to associated chat or other forum participation sessions and/or associated suggestible content that is likely to be an area of current focus for the respective STAN user or; due to the non-topic nodes also pointing to cross-associated topic nodes, the non-topic nodes can thereby indirectly point (by way of the intervening topic nodes) to associated chat or other forum participation sessions and/or associated suggestible content that is likely to be on-topic.
The types of raw CFi's (298 e?, 302 e?) or normalized/categorized CFi's (2980, 3020) that two or more STAN users have substantially in common are not limited to text-based information (textual CFi's). It could instead or additionally be musical or other sound-based information that has been normalized into a primitive that represents that non-textual information (see briefly the musical primitive object 30F.0 of FIG. 3F) and the users could be linked to one another based on substantial commonality of raw or categorized CFi's which are determined to be directed to substantially same primitives and/or substantially same or similar other points, nodes or subregions in music space rather than in a text-based space (e.g., topic space). The found commonality between STAN users can more generally be based on found substantially same focused-upon nodes and/or subregions in yet other nontextual spaces like a nontextual emotions space (where said other nontextual space can be a data-objects organizing space that uses a primitives data structure such as those of FIGS. 3F-3I, for example, in a primitives layer thereof and uses operator node objects (see FIG. 3Q) for defining more complex objects in, for example, emotion space in a manner similar to one that will be shortly explained for keyword expressions space). More specifically, two or more STAN users can be automatically joined online with one another based on substantial cross-correlation of shared emotion primitives, of shared sound primitives (see briefly FIG. 3G) and so on, as obtained from their respective CFi's; where the latter can be categorized as being textual CFi's or sound-related CFi's or emotions-related CFi's and so on. Alternatively or additionally, two or more STAN users can be automatically joined online with one another based on substantial cross-correlation of voice primitives (see briefly FIG. 3H) that are obtained from their respective CFi's. Alternatively or additionally, two or more STAN users can be automatically joined online with one another based on substantial cross-correlation of linguistic primitives (see briefly FIG. 3I) that are obtained from their respective CFi's. Alternatively or additionally, two or more STAN users can be automatically joined online with one another based on substantial cross-correlation of image primitives (see briefly FIG. 3M) that are obtained from their respective CFi's. Alternatively or additionally, two or more STAN users can be automatically joined online with one another based on substantial cross-correlation of body language primitives (see briefly FIG. 3N) that are obtained from their respective CFi's. Alternatively or additionally, two or more STAN users can be automatically joined online with one another based on substantial cross-correlation of physiological state primitives (see briefly FIG. 3O) that are obtained from their respective CFi's. Alternatively or additionally, two or more STAN users can be automatically joined online with one another based on substantial cross-correlation of chemical mixture objects defined by chemical mixture primitives (see briefly FIG. 3P) that are obtained from their respective CFi's.
Referring now to FIG. 3E, the more familiar among the Cognitive Attention Receiving Spaces, namely, the topic space mapping mechanism 313? is shown at the center of the diagram. For sake of example, other mapping mechanisms are shown to encircle the topic space hierarchical pyramid 313? and to cross link with nodes and/or subregions of the topic space hierarchical pyramid 313?. One of the other interlinked mapping mechanisms is a meta-tags data-objects organizing space 395. Although its apex-region primitives are not shown elsewhere in detail, the primitives of the meta-tags space 395 may include definitions of various HTML and/or XML meta-tag constructs which generally speaking, are a form of textual sequences or symbol strings whose symbols (codings) may include non-ASCII codes in addition to or as alternatives to ASCII coded symbols. CFi streamlets that include various combinations, permutations and/or sequences and/or chronological overlaps of meta-tag strings may be categorized by the machine system 410 on the basis of information that is logically linked to relevant ones of the nodes and/or subregions of the meta-tags space 395. More specifically, a meta-tag which indicates certain HTML content is to be highlighted by bolding, blinking, changing colors, etc. may logically link to representations of cognitions related to attention “getting” activities.
Yet another of the other interlinked mapping mechanisms shown in FIG. 3E is a keyword expressions space 370, where the latter space 370 is not illustrated merely as a pyramid, but rather the details of an apex portion and of further layers (wider and more away from the apex layers) of that keyword expressions space 370 are illustrated. Keyword expressions are another example of textual sequences or symbol strings whose symbols may include non-ASCII codes in addition to ASCII coded symbols, although typically they will include text strings (e.g., alphanumeric sequences). The “apex” layer or layers of the keyword expressions space 370 are also referred to herein as the primitive expressions clustering layer(s). More generally, for each of the cognition mapping mechanisms shown in FIGS. 3D-3E to be represented by an inverted pyramid, the at- or near-“apex” layer or layers may be referred to as the primitive expressions (or symbols or codings) clustering layer(s) of that mapping mechanism while the closer-to-base layers may be seen as containing clusterings of more complex representations of cognitions that build upon and build with the representations of more primitive cognitions representing “apex” layers. Representations which are clustered substantially close together (in a hierarchical and/or spatial sense) in a respective cognition mapping mechanism may be deemed to represent cognitions that are substantially same or similar to one another in a given kind of cognitive sense. Very briefly and as an example, say one primitive expression in keyword space 370 contains the symbols sequence, “Ab* Lincoln” where the asterisk is a wild card symbol such that Ab* can represent both of “Abraham” and “Abe”. Say as part of the brief example, another primitive expression contains the symbols sequence, “16th US President”. In one sense, both refer to the same persons and thus to the same cognitive sense, namely, that of Abraham Lincoln and he being the 16th US President. In one embodiment, the two symbol sequences, “Ab* Lincoln” and “16* U*S* President” would be clustered substantially close to one another in keyword space 370 (and/or in topic space) because they both may be deemed to represent respective cognitions that are substantially same or similar to one another in a given kind of cognitive sense. An example of a coded representation for a more complex cognition might be as follows: “(Ab* Lincoln) OR (16* U*S* President) AND (Civil War)”.
Before describing yet further details of the illustrated keyword expressions space 370, a quick return tour is provided here through the hierarchical, and plural tree branches-containing, structure (e.g., having the “A” tree, the “B” tree and the “C” tree intertwined with one another) of the topic space mechanism 313?. In the enlarged portion 313.51? of the space 313? as shown in FIG. 3E, a mid-layer topic node named, Tn62 (see also the enlarged view in FIG. 3X) resides on the “A” tree; and more specifically at a respective position along the horizontal branch number Bh(A)6.1 of the “A” tree but not on the “B” tree or on the “C” tree. Only topic nodes Tn81 and Tn51 of the exemplary hierarchy reside on the “C” tree. Topic node Tn51 is the immediate parent of node Tn62, and that parent links down to its child node, Tn62 by way of vertical connecting branch By(A)56.1 and horizontal connecting branch Bh(A)6.1. Other nodes (filled circle ones) hanging off of the “A” tree branch Bh(A)6.1 also reside on the “B” tree and hang off the latter tree's horizontal connecting branch Bh(B)6.1, where the B-tree branch is drawn as a dashed horizontal line in FIG. 3E.
Additionally, in FIG. 3E, topic node Tn61 is a parent to further children hanging down from, for example, “A” tree horizontal connecting branch Bh(A)7.11. One of those child nodes, Tn71, reflectively links to a so-called, operator node 374.1 in keyword space 370 by way of reflective logical link 370.6. Another of those child nodes, Tn74, reflectively links to another operator node 394.1 disposed in URL space 390 by way of reflective logical link 390.6. As a result, the second operator node 394.1 in URL space 390 is indirectly logically linked by way of sibling relationship on horizontal connecting branch Bh(A)7.11 to the first mentioned operator node 374.1 that resides in the keyword expressions space 370.
Parent node Tn51 of the magnified portion 313.51? of the topic space mapping mechanism 313? has a number of chat or other forum participation sessions (forum sessions) 30E.50 currently tethered to it either on a relatively strongly anchored basis (whereby a breaking off from, and drifting away from that mooring is relatively difficult) or on a relatively weak anchored basis (whereby a stretching away from, and/or a breaking off of the corresponding forum (e.g., chat room) and a drifting away from that mooring point Tn51 is relatively easier). Recall that members of chat rooms and/or other forums can vote to drift apart from one topic center (TC) and to more strongly attach one of their anchors (figuratively speaking) to a different topic centers as forum membership and circumstances change. In general, topic space 313? can be a constantly and robustly changing combination of interlinked topic nodes and/or topic subregions whose hierarchical organizations, names of nodes, governance bodies controlling the nodes, and so on can change over time to correspond with changing circumstances in the virtual and/or non-virtual world and the chat or other forum participation sessions attached to those plastic-wise re-configurable topic nodes or subregions can also change robustly.
The illustrated plurality of forum sessions, 30E.50 are servicing a first group of STAN users 30E.49, where those users are currently dropping their figurative anchors onto those forum sessions 30E.50 and thereby ‘touching’ topic node Tn51 to one extent of cast “heat” energy or another (e.g., casting attention giving energies on that node) depending on various “heat” generating attributes (e.g., duration of participation, degree of participation, emotions and levels thereof detected as being associated with the chat room participation and so on). Depending on the sizes and directional orientations of their halos, some of the first users 30E.49 may apply a halo-extended ‘touching’ heat to child node Tn61 or even to grandchildren of Tn51, such as topic node Tn71. Other STAN users 30E.48 may be simultaneously ‘touching’ other parts of topic space 313? and/or simultaneously ‘touching’ parts of one or more other spaces, where those touched other spaces are represented in FIG. 3E by pyramid symbol 30E.47. Representative pyramid symbol 30E.47 can represent keyword expressions space 370 or URL expressions space 390 or a hybrid keyword-URL expressions space (380) that contains illustrated node 384.1 or any other data-objects organizing space.
Referring to now to the specifics of the keyword expressions space 370 of the embodiment represented by FIG. 3E, a near-apex layer 371 of what in its case, would be illustrated as an upright pyramid structure, contains so-called, “regular” keyword expressions. An example of what may constitute such a “regular” keyword expression would be a string like, “??? patent*” where here, the suffix asterisk symbol (*) represents an any-length wildcard which can contain zero, one or more of any characters in a predefined symbols set while here, each of the prefixing question mark symbols (?) represents a zero or one character wide wildcard which can be substituted for by none or any one character in the predefined symbols set. Accordingly, if the predefined symbols set includes the letters, A-Z and various punctuation marks, the “regular” keyword expression, “???patent*” may define an automated match-finding query that can be satisfied by the machine system finding one or more of the following expressions: “patenting”, “patentable” “nonpatentable”, “un-patentable”, nonpatentability” and so on. Similarly, an exemplary “regular” keyword expression such as, “???obvi*” may define an automated match-finding query that can be satisfied by the machine system finding one or more of the following expressions: “nonobvious”, “obviated” and so on. The wildcard symbols need not be limited to these specific ones. In a later described data structure (see briefly 30W.0 of FIG. 3W) it will be seen how the definitions of what symbols serve as wild cards or not may be varied. A Boolean combination expression such as, “???patent*” AND “???obvi*” may therefore be satisfied by the machine system finding one or more expressions such as “patentably unobvious” and “patently nonobvious”. These are of course, merely examples and the specific codes used for representing wild cards, combinatorial operators and the like may vary from application to application. The “regular” keyword expression definers may include mandates for capitalization and/or other typographic configurations (e.g., underlined, bolded and/or other) of the one or more of the represented characters and/or for exclusion (e.g., via a minus sign) of certain subpermutations from the represented keywords.
In one embodiment, the “regular” keyword expressions of the near-apex layer 371 come to be spatially clustered around keystone expressions and/or are clustered according to Thesaurus-like senses of the words that are to be covered by the clustered keyword primitives. By way of example, assume again that a first node 371.1 in primitives layer 371 defines its keyword expression (Kw1) as “*lincoln*” where this would cover “Abe Lincoln”, “President Abraham Lincoln” and so on, but where this first node 371.1 is not intended to cover other contextual senses of the “*lincoln*” expression such as those that deal with the Lincoln™ brand of automobiles or the city of Lincoln, Nebr. Instead, the “*lincoln*” expression according to one of those other senses would be covered by another primitive node 371.5 that is clustered elsewhere (371.50) in addressable memory space near nodes (371.6) for yet other keyword expressions (e.g., Kw6?*) related to that alternate sense of “Lincoln”.