共查询到20条相似文献,搜索用时 15 毫秒
1.
Background
Animals that communicate by sound face the problem that the signals arriving at the receiver often are degraded and masked by noise. Frequency filters in the receiver''s auditory system may improve the signal-to-noise ratio (SNR) by excluding parts of the spectrum which are not occupied by the species-specific signals. This solution, however, is hardly amenable to species that produce broad band signals or have ears with broad frequency tuning. In mammals auditory filters exist that work in the temporal domain of amplitude modulations (AM). Do insects also use this type of filtering?Principal Findings
Combining behavioural and neurophysiological experiments we investigated whether AM filters may improve the recognition of masked communication signals in grasshoppers. The AM pattern of the sound, its envelope, is crucial for signal recognition in these animals. We degraded the species-specific song by adding random fluctuations to its envelope. Six noise bands were used that differed in their overlap with the spectral content of the song envelope. If AM filters contribute to reduced masking, signal recognition should depend on the degree of overlap between the song envelope spectrum and the noise spectra. Contrary to this prediction, the resistance against signal degradation was the same for five of six masker bands. Most remarkably, the band with the strongest frequency overlap to the natural song envelope (0–100 Hz) impaired acceptance of degraded signals the least. To assess the noise filter capacities of single auditory neurons, the changes of spike trains as a function of the masking level were assessed. Increasing levels of signal degradation in different frequency bands led to similar changes in the spike trains in most neurones.Conclusions
There is no indication that auditory neurones of grasshoppers are specialized to improve the SNR with respect to the pattern of amplitude modulations. 相似文献2.
The parallelisms in communicative behavior of domestic sheep and Northern fur seals within a herd are accompanied by parallelisms in parameters of sound signal, the calling scream. This signal ensures ties between babies and their mothers at a long distance. The basis of parallelisms is formed by amplitude modulation at two levels: the one being a direct amplitude modulation of the carrier frequency and the other--modulation of the carrier frequency oscillation. Parallelisms in the signal oscillatory process result in corresponding parallelisms in the structure of its frequency spectrum. 相似文献
3.
X Wang M B Sachs 《Philosophical transactions of the Royal Society of London. Series B, Biological sciences》1992,336(1278):399-402
We have investigated responses of the auditory nerve fibres (ANFS) and anteroventral cochlear nucleus (AVCN) units to narrowband 'single-formant' stimuli (SFSS). We found that low and medium spontaneous rate (SR) ANFS maintain greater amplitude modulation (AM) in their responses at high sound levels than do high SR units when sound level is considered in dB SPL. However, this partitioning of high and low SR units disappears if sound level is considered in dB relative to unit threshold. Stimuli with carrier frequencies away from unit best frequency (BF) were found to generate higher AM in responses at high sound levels than that observed even in most low and medium SR units for stimuli with carrier frequencies near BF. AVCN units were shown to have increased modulation depth in their responses when compared with high SR ANFS with similar BFS and to have increased or comparable modulation depth when compared with low SR ANFS. At sound levels where AM almost completely disappears in high SR ANFS, most AVCN units we studied still show significant AM in their responses. Using a dendritic model, we investigated possible mechanisms of enhanced AM in AVCN units, including the convergence of inputs from different SR groups of ANFS and a postsynaptic threshold mechanism in the soma. 相似文献
4.
Natural auditory stimuli are characterized by slow fluctuations in amplitude and frequency. However, the degree to which the neural responses to slow amplitude modulation (AM) and frequency modulation (FM) are capable of conveying independent time-varying information, particularly with respect to speech communication, is unclear. In the current electroencephalography (EEG) study, participants listened to amplitude- and frequency-modulated narrow-band noises with a 3-Hz modulation rate, and the resulting neural responses were compared. Spectral analyses revealed similar spectral amplitude peaks for AM and FM at the stimulation frequency (3 Hz), but amplitude at the second harmonic frequency (6 Hz) was much higher for FM than for AM. Moreover, the phase delay of neural responses with respect to the full-band stimulus envelope was shorter for FM than for AM. Finally, the critical analysis involved classification of single trials as being in response to either AM or FM based on either phase or amplitude information. Time-varying phase, but not amplitude, was sufficient to accurately classify AM and FM stimuli based on single-trial neural responses. Taken together, the current results support the dissociable nature of cortical signatures of slow AM and FM. These cortical signatures potentially provide an efficient means to dissect simultaneously communicated slow temporal and spectral information in acoustic communication signals. 相似文献
5.
Investigation of vibrational calling signals of about 500 species of small Auchenorrhyncha from Russia and adjacent territories has shown that more than 10% of the species studied produce signals fully or partially consisting of pure-tone components. Among these species, there are forms dwelling on various substrates including thick tree branches, slender twigs, grass stems, petioles, and leaves. Therefore, it is impossible to associate the presence of pure-tone signals in any species with the physical properties of the inhabited substrate. Pure-tone signals were recorded both in the largest and the smallest forms. Consequently, the type of signal frequency spectrum is not related to the insect size. Experiments under natural conditions confirm the assumption that pure-tone signals are more resistant to noise than wide-band ones. This property may compensate for the disadvantage of pure-tone signals arising from stronger attenuation in certain substrates. As a result, neither pure-tone nor noise signals give unambiguous advantages for vibrational communication. Since the carrier frequency of pure-tone signals of small Auchenorrhyncha increases with temperature, it seems to be determined by the contraction frequency of tymbal muscles, rather than by the resonance properties of any cavity or cuticular structure. Regular frequency modulations occur in the signals of many species; moreover, the calling signals of some species include both pure-tone and noise components. In most of the species studied, “frequency tuning” of their signals to the physical properties of a particular substrate is impossible because of the presence of frequency modulations, temperature-related variation in the carrier frequency, or the wide host range and the absence of strong preference for any particular plant organ. Sympatric species may differ in the carrier frequency of their pure-tone signals. If the signals occupy the same frequency range they sometimes differ in their frequency modulation pattern. Consequently, conspecific signal recognition not only by amplitude, but also by frequency pattern is possible in this case, which increases the efficiency of intraspecific communication. 相似文献
6.
Balzano Q 《Bioelectromagnetics》2003,24(7):483-488
The presence of harmonic products due to possible nonlinear interaction of amplitude modulated RF signals in living cells is best detected by using a cavity with high quality factor. Harmonic products generated by elementary oscillators can be trapped and accumulated in a cavity, permitting detection sensitivity much greater than in an open environment, where they would be radiated in all directions. The experimental method described herein is a systematic approach to detection of the non-Planck RF energy (if any) emitted by an exposed sample of living cells. Balzano and Sheppard [Balzano and Sheppard (2003): Bioelectromagnetics 24:473-482] classified the non-Planck RF emissions from living cells as coming from (1). nonlinear interactions and (2). inelastic interactions. Nonlinear harmonic products would appear in the band at twice the frequency of an amplitude modulated RF carrier. Inelastic interaction products resulting from the interaction between the incident RF energy and normally occurring mechanical vibrations are found in the band immediately adjacent to the carrier. Detection of the latter signals is difficult because of this close spectral proximity, for example, 1 part in 10(7) for 100 Hz modulation of a GHz carrier. Modern audio spectrum analyzers have excellent selectivity, providing 60 dB rejections only 2 kHz away from the carrier. By judicious selection of the amplitude modulation (AM) frequency, frequency of the RF carrier, and size of the biological sample, it is possible to achieve very high sensitivity (about -90 dBm) with commercially available instrumentation. The presence (or absence) of harmonics in the band adjacent to the amplitude modulated RF carrier would establish (or negate) the existence of coherent interactions between mechanical vibrations in the cell ensemble and the incident RF signal. 相似文献
7.
We experimentally demonstrated that tonal acoustic signals with a carrier frequency of 140–200 Hz had a repellent effect on male mosquitoes (Culicidae). Swarming males of Aedes diantaeus were concentrated in a small space near the auxiliary attracting sound source which simulated the flight sound of conspecific females (carrier frequency 280–320 Hz). Then, the resulting cluster of attracted mosquitoes was stimulated with test signals of variable amplitude and carrier frequency from a second loudspeaker. The direction of mosquito flight from the source of test sounds and a decrease in their number above the attracting sound source were used as the criteria of behavioral response. Pronounced avoidance responses (negative phonotaxis) of swarming mosquitoes were observed in the range of 140–200 Hz. Most of the mosquitoes left the area above the attracting sound source within one second after the onset of the test signal. Mosquitoes mostly flew up, sideways, and backwards in relation to the test acoustic vector. We presume that mosquitoes develop defensive behavior against attacking predatory insects based on analysis of auditory information. The range of negative phonotaxis is limited at higher frequencies by the spectrum of the flight sounds of conspecific females, and in the low frequency range, by the increasing level of atmospheric noise. 相似文献
8.
The airborne-sound and the vibratory signals produced by stridulating males, and the transmission of these signals in the natural biotope were investigated.The song of is composed of repeated uniform syllables with a rate of ca. 30/sec. Intensity approaches 100 dB SPL, 10 cm away from the animal. The spectrum shows three dominant frequency ranges around 8, 16 and 32 kHz.Airborne transmission of the song in such vegetation layers as are found in the biotopes of shows an excess attenuation which increases with frequency. The relative intensities of the frequency components of the song vary as a result of the kind of vegetation, the positions of emitter and receiver, and the separation distance. These relative differences in intensity may be useful during the phonotactic approach to conspecific partners, providing a measure of the distance from the sound source.Stridulating males also produce vibratory signals in the plants they sit on. The spectrum of these signals includes frequencies up to 8 kHz, the first dominant frequency of the song: low frequency components are induced in the plants via the legs and abdomen of the animal. The vibratory signals are transmitted mainly in the form of bending waves. Near the animal, amplitude modulation corresponds to that of the song. At greater distances, reflections and frequency-dependent propagation velocities, cause distortions of this time pattern. Transmission depends greatly on the mechanical properties of the particular plant, attenuation values of 20–50 dB/m being found. Nevertheless, in most cases, vibratory signals may be perceived up to 1.5 – 2 metres away from a stridulating male. 相似文献
9.
The accuracy of localizing the underwater sound source in the vertical-plane by the bottlenose dolphin was investigated using the method of instrumental conditioned reflexes with food reinforcement. The accuracy of determining the underwater sound in the vertical plane (the full angle) was on the average: 2 - 2,5 degrees for tonal signals with frequencies of 5, 20, and 120 kHz; pulsed clicks with the central frequency of 120 kHz and the exponential forms of amplitude alteration wavefronts were localized by the dolphin with an accuracy of 1,5 degrees. Among all marine mammals examined, dolphins are characterized by the maximal exact analysis of acoustic space. 相似文献
10.
J. Xu D. M. Gooler A. S. Feng 《Journal of comparative physiology. A, Neuroethology, sensory, neural, and behavioral physiology》1996,178(4):435-445
Single-unit recordings were made from 143 neurons in the frog (Rana p. pipiens) inferior colliculus (IC) to investigate how free-field sound direction influenced neural responses to sinusoidal-amplitude-modulated (SAM) tone and/or noise. Modulation transfer functions (MTFs) were derived from 3 to 5 sound directions within 180° of frontal field. Five classes of MTF were observed: low-pass, high-pass, band-pass, multi-pass, and all-pass. For 64% of IC neurons, the MTF class remained unchanged when sound direction was shifted from contralateral 90° to ipsilateral 90°. However, the MTFs of more than half of these neurons exhibited narrower bandwidths when the loudspeaker was shifted to ipsilateral azimuths. There was a decrease in the cut-off frequency for neurons possessing low-pass MTFs, an increase in cut-off frequency for neurons showing high-pass MTFs, or a reduction in the pass-band for neurons displaying bandpass MTFs. These results suggest that sound direction can influence amplitude modulation (AM) frequency tuning of single IC neurons.Since changes in periodicity of SAM tones alter both the temporal parameters of sounds as well as the sound spectrum, we examined whether directional effects on spectral selectivity play a role in shaping the observed direction-dependent AM selectivity. The directional influence on AM selectivity to both SAM tone and SAM noise was measured in 62 neurons in an attempt to gain some insight into the mechanisms that underlie directionally-induced changes in AM selectivity. Direction-dependent changes in the shapes of the tone and noise derived MTFs were different for the majority of IC neurons (55/62) tested. These data indicate that a spectrally-based and a temporally-based mechanism may be responsible for the observed results.Abbreviations
AM
amplitude modulation
-
CF
characteristic frequency
-
DI
direction index
-
FR
isointensity frequency response
-
GABA
gamma-aminobutyric acid
-
IC
inferior colliculus
-
ICc
central nucleus of the inferior colliculus
-
ITD
interaural time difference
-
MTF
modulation transfer function
-
PSTH
peri-stimulus time histogram
-
SAM
sinusoidal-amplitude-modulated
-
SC
synchronization coefficient
-
CN
cochlear nucleus 相似文献
11.
人类听觉的基本特性和机制与其他哺乳动物相似,因此,利用动物所作的听觉研究和获得的结果,有助于认识人类自身的听觉.围绕听觉中枢神经元对不同模式的声信号的识别和处理,简要综述了这方面的研究.声信号和声模式识别在听觉中枢对声信号的感受和加工中具有重要意义.听神经元作为声模式识别的结构和功能基础,对不同的声刺激模式产生不同反应,甚至是在同一声刺激模式下,改变其中的某个声参数,神经元的反应也会发生相应改变,而其反应的特性和机制均需要更多研究来解答.另外,声信号作为声信息的载体,不同的声信息寓于不同的声参数和声特征之中,研究发现,听觉中枢神经元存在相应的声信息甄别和选择的神经基础,能对动态变化的声频率、幅度和时程等进行反应和编码,并且,在不同种类动物上获得的研究结果极为相似,表明听觉中枢对不同声信号和声刺激模式的识别、分析和加工,具有共同性和普遍性. 相似文献
12.
In brown thrashers (Toxostoma rufum) and grey catbirds (Dumetella carolinensis) neither side of the syrinx has a consistently dominant role in song production. During song, the two sides operate independently, but in close cooperation with each other and with the respiratory muscles which are capable of adjusting expiratory effort to maintain a constant rate of syringeal airflow despite sudden changes in syringeal resistance. Phonation is frequently switched from one side of the syrinx to the other, both between syllables and within a syllable. When both sides of the syrinx produce sound simultaneously, their respective contributions are seldom harmonically related. The resulting “two-voice” syllables sometimes contain difference tones with prominent sinusoidal amplitude modulation (AM). Rarely, both sides simultaneously produce the same sound. In general, however, the frequency range of sound contributed by the right syrinx is higher than that of the left syrinx. The right syrinx is also primarily responsible for producing a rapid cyclical amplitude modulation which is a characteristic feature of some syllables. This kind of AM is generated by either repetitive brief bursts of sound from the right side that modulate the amplitude of a continuous sound arising on the left side or cyclically opening the right syrinx, allowing unmodulated expiratory air to bypass the phonating left side. 1994 John Wiley & Sons, Inc. 相似文献
13.
Lori WollermanR.Haven Wiley 《Animal behaviour》2002,63(1):15-22
Many animals communicate in environments with high levels of background noise. Although it is a fundamental prediction of signal detection theory that noise should reduce both detection and discrimination of signals, little is known about these effects in animal communication. Female treefrogs, Hyla ebraccata, in Costa Rica choose mates in large noisy multispecies choruses. We tested gravid females for preferences between computer-synthesized calls with carrier frequencies of 3240 and 2960 Hz (values near the mode and the fifth percentile of the population, respectively) in four levels of background noise from a natural chorus. In the absence of noise (signal/noise ratio >25 dB), females preferred the lower frequency. With moderate signal/noise ratios (6 and 9 dB), they did not discriminate between these frequencies. With low signal/noise ratios (3 dB), females preferred the frequency near the mode for the population. Similar experiments had previously demonstrated that females can detect the presence of a male's calls with signal/noise ratios of 3 dB or greater. Thus moderate levels of natural background sound reduced a female's ability to discriminate between males' calls even when she could detect them. In high levels of background sound, females abandoned discrimination for low-frequency calls and reverted to the task of detecting signals with modal properties for the population. These results justify recent theoretical analyses of the importance of receivers' errors in the evolution of communication. 相似文献
14.
Gabriele Micali Gerardo Aquino David M. Richards Robert G. Endres 《PLoS computational biology》2015,11(6)
Cells sense external concentrations and, via biochemical signaling, respond by regulating the expression of target proteins. Both in signaling networks and gene regulation there are two main mechanisms by which the concentration can be encoded internally: amplitude modulation (AM), where the absolute concentration of an internal signaling molecule encodes the stimulus, and frequency modulation (FM), where the period between successive bursts represents the stimulus. Although both mechanisms have been observed in biological systems, the question of when it is beneficial for cells to use either AM or FM is largely unanswered. Here, we first consider a simple model for a single receptor (or ion channel), which can either signal continuously whenever a ligand is bound, or produce a burst in signaling molecule upon receptor binding. We find that bursty signaling is more accurate than continuous signaling only for sufficiently fast dynamics. This suggests that modulation based on bursts may be more common in signaling networks than in gene regulation. We then extend our model to multiple receptors, where continuous and bursty signaling are equivalent to AM and FM respectively, finding that AM is always more accurate. This implies that the reason some cells use FM is related to factors other than accuracy, such as the ability to coordinate expression of multiple genes or to implement threshold crossing mechanisms. 相似文献
15.
Katrin Reimer 《Journal of comparative physiology. A, Neuroethology, sensory, neural, and behavioral physiology》1987,161(2):305-313
Summary Single neuron responses to sinusoidally amplitude modulated (SAM) signals were studied in the inferior colliculus of the horseshoe bat,Rhinolophus rouxi.57% of the neurons responded to SAM stimuli with periodical discharges synchronized to the modulation cycle. The proportion of cells driven by amplitude modulated signals was independent of the best frequency of the neurons. Best modulation frequencies were at or below 100 Hz in about 70% of the neurons. Synchronized activity could be elicited by modulation frequencies up to 400 Hz.Best SAM responses were observed at stimulus intensities 10 dB above threshold. Generally the BMF of a neuron did not change with intensity. The BMF decreased with decreasing modulation depth of the amplitude modulation.A trend for a topographical organization of neurons according to best modulation frequencies was detected. The results did not reveal any significant specialization of the bat's auditory system for coding of amplitude modulations as compared to other mammals.Abbreviations
BF
best frequency
-
BMF
best modulation frequency
-
CF
constant frequency
-
FM
frequency modulation
-
IC
inferior colliculus
-
SAM
sinusoidal amplitude modulation
-
SFM
sinusoidal frequency modulation 相似文献
16.
Acoustic Communication in a Black‐Headed Gull Colony: How Do Chicks Identify Their Parents? 总被引:2,自引:0,他引:2
Isabelle Charrier Nicolas Mathevon Pierre Jouventin & Thierry Aubin 《Ethology : formerly Zeitschrift fur Tierpsychologie》2001,107(11):961-974
Chicks perform conspicuous begging behaviour in response to the arrival of a parent. In seabirds colonies, as nests are close to each other, chicks are permanently surrounded by sound and visual stimuli produced by adult conspecifics approaching their nests. However, in spite of these conditions, black-headed gull chicks begin to vocalize as their parent approaches even before they can see it. In this paper, we report field experiments testing sound-based discrimination of parents by black-headed gull chicks. Focusing on the 'long call', i.e. the signal emitted by parents when coming back to the nest, we investigate here the acoustic parameters used for this recognition process. By playback experiments using modified 'long calls', we demonstrated that signals without amplitude modulation still elicit responses in chicks. In contrast, frequency modulation appears essential. In the frequency domain, experiments revealed that frequency analysis is precise. Chicks did not react when the frequency spectrum of parental call was shifted 20 Hz down or up. The totality of harmonics is not necessary: chicks require only two harmonics to discriminate between parents. Signal redundancy is of great significance since a minimum of four successive syllables in parental 'long call' are required to elicit reaction in the chick. 相似文献
17.
C.I. Christodoulou P.A. Kaplanis V. Murray M.S. Pattichis C.S. Pattichis T. Kyriakides 《Biomedical signal processing and control》2012,7(3):265-269
In this work, multi-scale amplitude modulation–frequency modulation (AM–FM) features are extracted from surface electromyographic (SEMG) signals and they are used for the classification of neuromuscular disorders. The method is validated on SEMG signals recorded from a total of 40 subjects: 20 normal and 20 abnormal cases (11 myopathy, and 9 neuropathy cases), at 10%, 30%, 50%, 70% and 100% of maximum voluntary contraction (MVC), from the biceps brachii muscle. For the classification, three classifiers are used: (i) the statistical K-nearest neighbor (KNN), (ii) the self-organizing map (SOM) and (iii) the support vector machine (SVM). For all classifiers, the leave-one-out methodology is used to validate the classification of the SEMG signals into normal or abnormal (myopathy or neuropathy). A classification success rate of 78% for the AM–FM features and SVM models was achieved. These results also show that SEMG can be used as a non-invasive alternative to needle EMG for differentiating between normal and abnormal (myopathy, or neuropathy) cases. 相似文献
18.
JEAN-PIERRE RICHARD 《Bioacoustics.》2013,22(1):45-60
ABSTRACT This note presents a program for analysis and synthesis of sounds designed to be used with an Amiga micro-computer. The user is in real-time control of the capture of the signal with reference to both time and amplitude parameters. Subsequently the chosen signal can be edited and analysed. Sonograms can be obtained at three resolutions, both in time and frequency. Measurements of time can be made on the signal, of frequency on the sonogram and of amplitude on sections. Three functions for the extraction of the fundamental are available. A synthetic sound may be obtained, either by modifying the amplitude and frequency modulation of an existing sound, or by creating a new sound by additive synthesis. All the commands can be selected by using a mouse, and the program is user-friendly. 相似文献
19.
Curtis J. Condon Su-Hua Chang Albert S. Feng 《Journal of comparative physiology. A, Neuroethology, sensory, neural, and behavioral physiology》1991,168(6):709-725
Summary Response characteristics of 130 single neurons in the superior olivary nucleus of the northern leopard frog (Rana pipiens pipiens) were examined to determine their selectivity to various behaviorally relevant temporal parameters [rise-fall time, duration, and amplitude modulation (AM) rate of acoustic signals. Response functions were constructed with respect to each of these variables. Neurons with different temporal firing patterns such as tonic, phasic or phasic-burst firing patterns, participated in time domain analysis in specific manners. Phasic neurons manifested preferences for signals with short rise-fall times, thus possessing low-pass response functions with respect to this stimulus parameter; conversely, tonic and phasic-burst units were non-selective and possessed all-pass response functions. A distinction between temporal firing patterns was also observed for duration coding. Whereas phasic units showed no change in the mean spike count with a change in stimulus duration (i.e., all-pass duration response functions), tonic and phasic-burst units gave higher mean spike counts with an increase in stimulus duration (i.e., primary-like high-pass response functions). Phasic units manifested greater response selectivity for AM rate than did tonic or phasic-burst units, and many phasic units were tuned to a narrow range of modulation rates (i.e., band-pass). The results suggest that SON neurons play an important role in the processing of complex acoustic patterns; they perform extensive computations on AM rate as well as other temporal parameters of complex sounds. Moreover, the response selectivities for rise-fall time, duration, and AM rate could often be shown to contribute to the differential responses to complex synthetic and natural sounds.Abbreviations
SON
superior olivary nucleus
-
DMN
dorsal medullary nucleus
-
TS
torus semicircularis
-
FTC
frequency threshold curve
-
BF
best excitatory frequency
-
PAM
pulsatile amplitude modulation
-
SAM
sinusoidal amplitude modulation
-
SQAM
square-wave amplitude modulation
-
MTF
modulation transfer function
-
PSTH
peri-stimulus time histogram 相似文献
20.
Natural auditory scenes such as frog choruses consist of multiple sound sources (i.e., individual vocalizing males) producing sounds that overlap extensively in time and spectrum, often in the presence of other biotic and abiotic background noise. Detection of a signal in such environments is challenging, but it is facilitated when the noise shares common amplitude modulations across a wide frequency range, due to a phenomenon called comodulation masking release (CMR). Here, we examined how properties of the background noise, such as its bandwidth and amplitude modulation, influence the detection threshold of a target sound (pulsed amplitude modulated tones) by single neurons in the frog auditory midbrain. We found that for both modulated and unmodulated masking noise, masking was generally stronger with increasing bandwidth, but it was weakened for the widest bandwidths. Masking was less for modulated noise than for unmodulated noise for all bandwidths. However, responses were heterogeneous, and only for a subpopulation of neurons the detection of the probe was facilitated when the bandwidth of the modulated masker was increased beyond a certain bandwidth - such neurons might contribute to CMR. We observed evidence that suggests that the dips in the noise amplitude are exploited by TS neurons, and observed strong responses to target signals occurring during such dips. However, the interactions between the probe and masker responses were nonlinear, and other mechanisms, e.g., selective suppression of the response to the noise, may also be involved in the masking release. 相似文献