SYSTEM AND METHOD FOR ADAPTIVE RSVP-BASED TEXT DISPLAY
BACKGROUND OF THE INVENTION
Technical Field of the Invention
The present invention relates to rapid serial visual presentation (RSVP) , and particularly to improvements of the method that accounts for text characteristics. Background of the Invention
There is a rapid increase of the use of portable tools such as mobile telephones and personal digital computers
(PDC) . These tools are characterized, among other things, of having a fairly limited display unit for display of text and graphics. The type of applications commonly executed on these tools usually have a simple user interface involving, e.g., selection of items from a list, key input of short messages, and reading of reasonably short messages. However, the evolving mobile computing community will radically change the scene whereby users connect to the
Internet for reading on line or downloading into the portable device large quantities of data. Exemplary, in a mobile office environment people will access extensive documents for reading and editing and the entertainment industry will offer electronic books. Thus, the problem of presenting large volumes of information for reading on a small display unit is becoming of increasing interest.
Reading text, such as an electronic book, from a conventionally sized computer monitor involves much of the same activity as in reading printed material. Several lines of text are simultaneously presented on the computer monitor and the reader reads each line of text before scrolling or paging to the next set of lines of text.
A problem exists, however, when attempting to read a relatively large amount of text from a relatively small display, such as a display screen for a PDC or Smartphone. In this case, the reader cannot efficiently read a relatively large amount of text by utilizing the scrolling and other features utilized in reading from a large computer screen.
Another problem is related to difficulties caused by a reading disability. People suffering from certain reading disabilities have difficulties processing texts comprising several lines. Under these circumstances, it may be advantageous to present the text according to the RSVP paradigm in smaller sections. The presentation may be made on any display whereby, of course, a large display will be only partially exploited.
Description of related art A display method is disclosed in the US patent application US 4,845,645 (E. Martin and K. R. Boff) characterized by
the sequential presentation of independent frames of information in one small display window on an electronic display device, the display window being one spatial location which may be viewed in its entirety without scanning eye movements (saccades) . However, whereas Martin and Boff is concerned with the display of real-time visual information, there is no mention of applicability to reading text and to aspects related to reading characteristics . Generally speaking, Rapid Serial Visual Presentation, or RSVP, involves the brief display of text segments at a fixed location. Typically, the reader sets the rate of presentation per segment, and each segment is shown for an equivalent amount of time. Segmenting of text is generally based on either the linguistic structure or the number of characters displayed. Segmenting according to linguistic structure appears to be more efficient than the segmentation based on quantitative factors. In most of the research with RSVP, the display duration of each segment is very brief, as implied by the word "rapid" in the acronym. Regardless of the rate of presentation, the functional relevance of RSVP lies in the fact that a small amount of text, such as a word or small number of words is presented to the reader at one time. This allows for use of small
screens, like screens used in PDCs or wireless telephones. By presenting text as a stream of text to a reader such that only a small amount of text is presented to the reader at a time, users of PDCs, wireless telephones and other devices may easily and efficiently read relatively large amount of text.
"Teenax Software Engineering" (available at http://eet.com/news/97/958news/aided.html) has released a machine-assisted reading software that segments sentences into a variable-speed stream of single words according to the RSVP paradigm.
Reading efficiency may be defined to depend on reading speed and comprehension. Thus, a protocol for presenting text efficiently on a small display should include factors known to affect comprehension as well as factors known to affect reading speed. However, the RSVP method, as conventionally applied, fails to consider some facts known to affect reading speed and reading comprehension. For example, reading speed will vary depending on the difficulty of the text. The internal comprehension process of the reader will affect the time to mentally process a completed sentence and, thus, when a new sentence may be commenced. A difficult sentence will require a longer mental processing time. Various methods are known for
calculation of the difficulty of a text, e.g. the LIX- method (Bjδrnsson, C.H., 1958, Stockholm, Liber) or the Flesch-method (Talburt, J., ACM, 1986, 0-89791-186- 5/86/0600/0114) . Just and Carpenter. (Psychological Review, Vol. 8, No. 4; July 1980) have presented a model of reading comprehension that accounts for the allocation of eye fixations when reading text from paper. Just and Carpenter found, e.g., that the gaze duration was longer, i.e. the mental processing load was greater, when processing infrequent words, when integrating information from important clauses, and when making inferences at the end of a sentence. Thus, it would be a benefit to improve the RSVP method to account for factors, which affect reading speed and comprehension such as to improve the efficiency in reading large amounts of text from a small display.
SUMMARY OF THE INVENTION
It is an object of the present invention to provide a method, system and computer readable medium that overcomes shortcomings in prior computer-aided reading systems and solves a significant need for such a system by improving efficiency in reading large text volumes from a small display. According to a preferred embodiment of the present
invention, a reading file is divided into text segments comprising a few words. Each text segment is presented on the same visual location of a display unit. After each sentence delimiter, e.g. period, colon, semi-colon, question mark, exclamation mark, a blank window is inserted of duration, which depends on the text characteristics of the sentence. A text segment is exposed for a time duration, which depends on the text characteristics of the segment. Segment exposure times and duration of blank windows are calculated such that a predetermined mean reading speed is obtained. Said same visual location comprises at least a part of at least one display line.
Thus, the present invention accounts for the fact that reading speed is not constant throughout a volume of text but varies as a function of the characteristics of the text. Further, the invention accounts for the fact that the comprehension of a sentence requires a certain processing time, which depends on the characteristics of the sentence. In a preferred embodiment of the invention, a first index is calculated for each text segment that affects the exposure time of that segment and a second index is calculated that affects the time of a blank window inserted after a sentence delimiter. The model by Just and Carpenter
may be the basis for the calculations of segment exposure time and duration of a blank window.
Thus, it is an object of the present invention to disclose an adapted RSVP- ethod characterized in that text segments are displayed for a time duration adapted to text characteristics and specified reading speed and further in that a blank window, of duration adapted to text characteristics and specified reading speed, is inserted after a sentence delimiter. Another object of the present invention is to disclose a system for display of segmented text on an electronic display device, the system comprising a display device, such as a portable, hand-held communication device having a small display, a processing unit and a memory unit. Said processing unit implementing software for display control and said memory unit storing a list of words and their frequency. The system further comprising a text-analyzing unit, a storage unit storing words and their frequency, an output unit creating segmented text and associated control information.
A further object of the present invention is to provide an adapted RSVP-method characterized in that text segment display time is scaled to fit a new reading speed, resulting from an instantaneous change of the current
reading speed, such that complex segments are scaled differently from less complex segments .
By text we shall, in relation to the present invention, understand any printed code that conveys a meaningful message when interpreted by a human reader. Thus, e.g., Chinese characters are included within the scope of the invention.
These objectives are obtained by a system and method as set out in the appended claims. Further scope of applicability of the present invention will become apparent from the detailed description given hereinafter. However, it should be understood that the detailed description and specific examples, while indicating preferred embodiments of the invention, are given by way of illustration only, since various changes and modifications within the scope of the invention will become apparent to those skilled in the art from this detailed description.
BRIEF DESCRIPTION OF THE DRAWINGS
A more complete understanding of the system and method of the present invention may be obtained by reference to the following Detailed Description when taken in conjunction with the accompanying drawings, which are given by way of illustration only, and thus are not limitative of the present invention and wherein: Figure 1 is an arrangement, according to a preferred embodiment of the present invention, for downloading to a mobile terminal of segmented text prepared by a reading service accessed through a communications network. Figure 2 is a block diagram, according to a preferred embodiment of the present invention, illustrating entities for analyzing and preparing an output file of segmented text.
Figure 3 is a block diagram of an arrangement at a mobile terminal, according to a preferred embodiment of the present invention, for processing and presentation of segmented text on a display unit.
Figure 4 is an exemplary format of an output file comprising a text segment and control information.
Figure 5 is an exemplary format of a list of words and word frequency.
Figure 6 is an exemplary format of a list of words, associated with a particular user, including word frequency and timer information.
Figure 7 is a flow chart illustrating the steps in the processing of a selected text for display on a display unit .
Figure 8 is a flow chart illustrating in more detail the steps in the analysis of a text. Figure 9 is an exemplary diagram illustrating the scaling of exposure time at adjustment of reading speed.
Figure 10 is a flow chart illustrating the analysis of a text with respect to word frequency.
Figure 11 is a flow chart illustrating the management of a timer related to infrequent words .
DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENT The present invention will now be described more fully hereinafter with reference to the accompanying drawings in which a preferred embodiment of the invention is shown. Referring to Figure 1, a mobile terminal 10 establishes a connection with a mobile network 20 through a base station (BS) 21. The mobile terminal 10 may, e.g., be a mobile telephone, a Personal Digital Assistant (PDA) or any other device or set of connected devices having radio access to a
mobile network. Said connection may utilize the WAP protocol (Wireless Application Protocol) , the GPRS protocol (General Packet Radio Service) or any other communication protocol for transfer of data. If access is desired to a service application in a data network 30, e.g. the
Internet, the communication proceeds through a gateway server (GW) 22 and further to a service control 31 implementing the desired service. Exemplary, the desired service is a reading service in which case the user of mobile terminal 10 is presented with a list for selection of a text reading file. In response to selecting a reading file and indicating a desired mean reading speed, service control 31 accesses the selected file from a database source 32. The service control 31 performs a text analysis of the selected reading file and creates an output file of data segments comprising a text segment part and a control information part. In preparing an output file the service control 31 may also use data, provided by the terminal at session initiation, pertaining to display characteristics of mobile terminal 10, e.g. line length of display, and font types supported by the terminal. The output file is, thereafter, downloaded into the mobile terminal 10 and processed for presentation on the terminal display whereby said control information is used by a display control unit
to control presentation characteristics. In a preferred embodiment, said presentation characteristics comprise the time of exposure of a text segment and the time of duration of a blank window separating two consecutive sentences. Figure 2 illustrates in more detail an exemplary structure of the service control part and its communication links to other entities. At 10 a service control part is shown comprising a communications control 11, an analysis control 12, and an output unit 13. Said units 11, 12, and 13 communicate through an internal communications link 14. A text database 21 provides the system with selected text files. Text analysis, performed by analysis control 12, may require access, through communications control 11, to a global list of infrequent words 22 and to a user associated infrequent word list 23. The database entities 21, 22 and 23 may, e.g., be addressed by a URL (Universal Resource Locator) . Whereas addresses for location of text database 21 and global word list 22 may be pre-stored at analysis control 12 the user at session initiation may provide the address locating a user word list 23. An output file, comprising a list of segmented text and control information, is stored at the output unit 13. The output file may then, through communications control 11 and gateway 30, be routed to a mobile network for download into
a mobile terminal. A person skilled in the art would foresee the possibility of locating service control 10 or user word list 23 or both these entities at a mobile terminal . Figure 7 shows a flow chart of the process just described. At Si the mobile terminal connects, through a mobile communication network, to a text service control. At step S2 basic data required for the subsequent analysis step, such as display characteristics, are transferred to the service control. Step S3 involves identifying and accessing the storage location of a user associated wordlist of infrequent words . The location may exemplary be defined by means of a URL. However, it should be apparent that said word list may also be stored at a storage device in the mobile device or any other accompanying device connected to the terminal by wire or short range radio such as Bluetooth™. The user selects at S4, from a list of selections, a text for reading. In response to the selection of text the service control retrieves the text from a text database. A desired initial mean reading speed, set by the user, is transferred at step S5. A reading speed may, e.g., be set by the user at a mobile device by operating a button or by selection of a suitable user menu and typing or otherwise selecting a desired value of the
initial reading speed. The user may, during the course of reading a text, change the value of reading speed in a similar way. The service control, at step S6, determines content area and language of the selected text and accesses a general list of infrequent words specific for the area and language. A general list of infrequent words may be stored at the service control or otherwise accessible through an address to a network storage location. At step S6, the service control executes a program for text analysis of the selected text. The analysis is exemplary described in more detail by reference to Figure 8. At step Si, the next sentence to be processed is identified. Text segmentation is performed at step S2. Segmentation is most effectively based on linguistic structure, e.g. by distinguishing noun phrase and verb phrase. At step S3 each resulting segment is analyzed and at least a segment parameter is determined. The analysis may be based on the model of Just and Carpenter. Said at least a segment parameter is indicative of the time of exposure of that segment. Moving now to step S4 the exposure time of the segment is calculated.
Just and Carpenter use the following formula for calculation of the segment exposure time when reading text from paper:
Sj = Σ bn *znj + a, ( 1 )
Where Sj is the exposure time of segment j and Znj are segment parameters that code a number of properties of segment j, such as for example length of segment, word normative frequencies, number of novel words. Coefficient bn is obtained from an adaptation of the model to experimental results. Normative word frequencies are obtained from a predefined table reflecting the result of an analysis of a large number of texts within the particular text field and language. Just and Carpenter publishes values for bn corresponding to a typical mean reading speed. It is foreseen that the adaptation of equation (1) to other reading rates will result in separate sets of bn. Reading rates, for which there is no corresponding set of bn, may be incorporated by extrapolation. In step S5, it is determined if the current segment is last in sentence or paragraph. If not, the routine proceeds to analyze the next segment in step S3. If the segment is last in sentence or paragraph, the routine proceeds to step S6 to determine the duration of a blank window following after the end of sentence or paragraph. The length of duration of a blank window is calculated, e.g., using the formula of Just and Carpenter
∑bn *Znj where the parameter Z now codes that the sentence is last in sentence or last in paragraph as is the case. The
factor b is the corresponding experimental weight in milliseconds .
In step S7 it is determined if the sentence is last in text in which case the routine ends, or otherwise the routine repeats step SI selecting the next sentence.
Returning now to Figure 7, the result from the analysis step S6 is further processed in step S7. In particular, step S7 involves the creation of an output file comprising at least a text segment and thereto attached control information. Said control information relates to the control of segment exposure time and time duration of a blank window following after a sentence delimiter. In addition, step S7 involves scaling said time values such that the total sum of segment exposure times and time duration of blank windows summed over all segments and sentences corresponds to the initial mean reading speed set in step S5. Figure 4 illustrates exemplary the format of a record in the output file in step S7. It is apparent for a skilled person that the control information may include other information in addition, e.g. formatting information such as font size and font color.
The step of processing a downloaded file for presentation on a terminal display device is described in more detail with reference to Figure 3. A text memory 10 stores the
downloaded file. Control unit 20 comprises processing means and outputs a first control signal 21 commanding a readout from memory 10 of a data segment. The control unit separates from the data segment at least part of said control information, which is routed to a display control unit 30, and text information, which is routed to a display input buffer 40. The control unit formats the text, routed to the buffer 20, according to at least part of said control information. In response to at least a second control signal from the control unit, the text stored in buffer 20 is displayed on the display device 50 under control of display control 30.
In response to the elapse of the appropriate time for exposure of a text display window and a possible blank window, the control unit 20 again directs at least a first control signal to initiate readout of the next data segment from the text memory 10. The process is repeated until there is no more text segments or the process stops due to manual interaction or to detection of an error.
DESCRIPTION OF ALTERNATIVE EMBODIMENTS
The blank window duration also referred to as sentence wrap up period, and determined in Figure 8, step S6, is usually a constant in prior art systems. In an alternative
embodiment of the present invention the duration of the blank window is made dependent on the reading difficulty calculated for the complete text. Thus, the blank window is made longer for a difficult text and is shortened for a more easy text. From an analysis of a large number of texts, in the same field and language, a mean reading difficulty and corresponding normal blank window duration are determined. According to the alternative embodiment, said scaling is proportional to the deviation of the current value of the reading difficulty from a mean reading difficulty value. A reading difficulty may be determined, e.g., by a modification of the original LIX-method according to the equation: Mean sentence length + %long words + %infrequent words (Eq 2) .
Mean sentence length is expressed as number of words and obtained by dividing the total number of words in the text by the number of sentences. A long word is defined as a word having at least a certain number of characters. A word is infrequent if its frequency, as obtained from a frequency analysis of texts in the same field and language, does not exceed a certain value.
It is noted that the creation of an output file of text segments is the result of calculations based on a certain
initial reading speed. For each selected reading speed there is a corresponding set of coefficients bn. For practical reasons only a limited number of distinct sets of coefficients is possible. Therefore, for other reading speeds, extrapolation is necessary. Further, in order to be able to change the reading speed without redoing the calculation of coefficient bn and downloading a revised text file extrapolation can be utilized. Thus, in another embodiment of the present invention, segment exposure time and duration of blank window are extrapolated by scaling by factor 1/k to satisfy a deviation of reading speed by a factor k from an original value.
In a further embodiment, in order to satisfy a specified new reading speed, segment exposure time is scaled with factor 1/k if the new reading speed is lower than the initial reading speed. If the new reading speed is higher than the original reading speed, a scaling factor is used such that a text segment with a value of the original exposure time exceeding a cutoff value is relatively less scaled than a segment with a value of the original exposure time below said cutoff value. Figure 9 illustrates the case of an exposure time SI exceeding a cutoff value SCutoff respectively an exposure time S2 below said cutoff value. Curve I illustrates the scaling of Sj using the factor 1/k
where k is the factor by which the reading speed has changed, i.e. a doubling of the reading speed (k=2) results in reducing to half the exposure time. Curve II illustrates the scaling according to the modified scheme. It can be seen that for the high value SI curve II is above curve I, i.e. the decrease of exposure time is slower than according to curve I .
In still another embodiment, the effect of word repetition is accounted for. According to this embodiment, a list of infrequent words is maintained for each individual user. When an infrequent word is encountered, said list is consulted and an initial value of the frequency of said word is read from the list. If the initial value exceeds a certain value, it is used to calculate a reduced effect on the exposure time of the segment containing the word. For each new occurrence of the word within the text currently being read, the frequency value is stepped a certain value. Thus, at the next occurrence of the word, its effect on the exposure time is still less. If the frequency value reaches a certain high value, it will be considered to be an ordinary word and marked as ordinary in the infrequent word list. At the next occurrence of the word, it will contribute to segment exposure time as an ordinary word. On the other hand, after finishing reading of the text, the
frequency value, for words not marked as ordinary, will be gradually decreased according to a certain time scale. Thus, if a word has not been in use for a long time, its contribution to segment exposure time at the next occurrence may be higher than at the last occurrence. The flow charts Figure 10 and Figure 11 illustrate the steps taken by the text service control when analyzing a text with respect to infrequent words. The steps according to Figure 10 are included in the analyzing phase at step S6 in Figure 7. Turning now to Figure 10, the process starts by accessing the next word in the current text at step Si. At step S2 a check is made if the selected word is considered infrequent by accessing a list of words pertaining to the subject matter field and text language. If the word is not in the list of infrequent words the method returns to step SI and selects the next word. In contrary, the method proceeds to step S4 and determines, through access to a user-associated list of infrequent words, the current frequency of the word. If the word is not in the list, it is added to the list and the frequency value is stepped. In step S7 a timer is initiated. Otherwise, if the word is already in the list, the method route to step S8 and the current frequency is read from the list. In addition, the frequency is stepped. In step S9, the word timer is reset.
In step S10 it is determined if word frequency now exceeds a certain limit value and, if that is the case, the word is marked as ordinary in the list. Otherwise, the method proceeds to step Sll where the contribution of the word to segment exposure time is calculated. In step S12 the method either route to step SI to select the next word or, if the text analysis is complete terminates.
Turning now to Figure 11, the flow chart illustrates steps to control the frequency value of words in the user- associated list of infrequent words. The steps, according to Figure 11, may either be executed by the text service logic or by program logic implemented in the device storing the list. In step Si, the next word in the list is identified. In step S2 the timer is stepped. In step S3 it is determined if a preset time has elapsed since the word was last used. If said time has not elapsed, the method returns to step Si to read the next word in the list. Otherwise, the method proceeds to step S4 where the word frequency is decreased. In step S5 it is determined if the list has been completed in which case the method stops, or otherwise returns to step SI to read the next word. In one embodiment of the present invention, the amount by which word frequency is decreased in step S4 increases with time.
The invention being thus described, it will be obvious that the same may be varied in many ways. Such variations are not to be regarded as a departure from the spirit and scope of the invention, and all such modifications as would be obvious to one skilled in the art are intended to be included within the scope of the following claims.